Monday, April 18, 2011

Liberals and NDP make big gains

A flurry of polls were released this morning (Nanos, Angus-Reid, and Léger Marketing). And with the incorporation of all of the polls that were released over the weekend (EKOS, Nanos, Forum), there's a lot of change in today's projection - and it isn't good news for Stephen Harper.
Since Friday, the Conservatives have dropped 0.3 points to 38.7% and three seats to 149, the lowest they've stood in this campaign so far. The Liberals have dropped 0.2 points to 28%, but because of gains in Ontario, Quebec, and Atlantic Canada the party is up five seats to 78. That's a gain of one seat since the government fell.

The New Democrats were buoyed today by some great polls for them from online pollsters Léger and Angus-Reid, and they are up 0.8 points in the projection to 17.6%. Long-time readers will know that is a huge update-to-update change at the national level. They are also up one seat to 34.

I was not able to get my hands on the full details of the Léger poll for this update, and media reports by the QMI Agency were incomplete. Only the national and Quebec numbers were added to the projection. Léger is usually pretty quick with updating their site, so hopefully I will be able to add the other regional numbers to the projection tomorrow.

The Bloc Québécois is down 0.3 points nationally to only 8.6%, and they are down three seats to 47. The Greens are down 0.2 points to 5.8%.

The Conservatives have remained relatively stable west of Quebec, dropping 0.2 points in Ontario but gaining 0.1 points in British Columbia and the Prairies and 0.2 points in Alberta.

However, they are down 0.4 points to 20.3% in Quebec, and are down a big 1.2 points in Atlantic Canada.

The Liberals are up marginally there and in Ontario, but dropped a full point in the Prairies. The bright spot for them is their closing of the gap in Ontario by 0.3 points and their holding steady in Quebec while the Bloc dropped a point.

The New Democrats, however, are up everywhere. British Columbia saw only a modest gain, but the party jumped 0.4 points in Ontario, 0.6 points in the Prairies, 0.7 points in Atlantic Canada, 0.8 points in Alberta, and an amazing 1.7 points in Quebec. They now stand at 18% in the province, within reach of the Liberals and Conservatives.

And that is a problem for the Bloc, which has dropped another point to only 35.9% support.

There have been quite a few seat changes since Friday morning.

In Ontario, the Liberals have retaken Brampton - Springdale from the Conservatives. Ruby Dhalla, the incumbent Liberal candidate, is again the favourite.

In Atlantic Canada, the Liberals have moved ahead in the projection in Random - Burin - St. George's, a riding in Newfoundland & Labrador. Judy Foote is the Liberal incumbent there. And in Prince Edward Island, Guy Gallant is now favoured over Gail Shea in the riding of Egmont. That's another Liberal pick-up from the Tories.

The Liberals have taken Ahuntsic (Noushig Eloyan) and Brossard - La Prairie (Alexandra Mendes) from the Bloc Québécois in and around Montreal. The New Democrats have also taken Gatineau from the Bloc. Françoise Boivin, the NDP candidate, is the favourite here.

That makes for a lot of change, and it appears it could continue. The NDP is riding high in Quebec and though they are still a little ways away from taking a third seat, it is certainly a good possibility that they will get to that point. But the kinds of support we saw in today's polls will need to hold firm.

The Conservatives are looking flat, but they are still comfortably ahead. But their prospects for a majority seem to be slipping away.


  1. A large number of polls, and a large volume of change - real or polling accident? Do we know if a small number of polls are responsible for most of the changes? (i.e. what happens if you add all but one poll?)

  2. The projection model doesn't swing wildly, which is why I don't have the NDP at over 20% in Quebec yet.

    As an example, if I remove the Angus-Reid poll from the projection, which had the NDP at 26% in Quebec, the party loses 0.4 points in the projection in that province.

  3. You do great work on this site, but some of your conclusions seem wrong. No one seriously thinks that CPC fortunes are "the lowest they've stood in the campaign so far". Most reasonable people know that PM Harper is heading for a majority government, and that the movement among the other parties is just part of that realization by anti-Harper voters. There will be further ups and downs as voters think about strategic votes. The fact remains, however, the the Conservative campaign is on cruise control, which means their polling is telling them they're on the right track. Good luck with your site.


  4. WildeMan,

    That certainly is possible. I was referring to 149 seats being the lowest I've projected them to be at so far in this campaign.

  5. That's interesting that these changes cause a loss of CPC seats to Liberals and NDP, rather than vote splitting giving more CPC seats. I hope your model is right and that we will see a lot more of this in the next 2 weeks and particularly on May 2nd.

  6. Thanks for this most excellent website! I suspect NDP gains in Quebec will come mainly in Montreal at the expense of the Liberals. This may not be reflected in your seat projection model.

  7. The NDP seems to be gaining a lot in Québec, but I'm wondering if this is crimping your model's style.

    I know that, typically, prediction models are based on past results and are valid as long as movements remain uniform in regions, but when large swings occur, they rarely do so uniformly. The NDP in Québec seems to have become subject to this, it has never been so high in the polls, we don't know where they are going to be getting their votes, if it will be concentrated in a few riding, giving them many seats, or if it is uniform, which will not yield them many seats at all.

    An example of that is the Conservatives in Québec. When they had that breakthrough in the Québec city area, their previous results had been very low, but if their province-wide gains had been uniform, they wouldn't have won as many seats, it's because their vote gains were concentrated in a specific region that they got as many seats as they did.

    I know there is no real way to get around that problem, but maybe it'd be best to mention these limits in further predictions if you think my point is accurate.

  8. Hey Éric,

    Would you be open to putting the Bloc seats and national votes between the Conservative and Liberals as well? Just think it'd look better.

    For the curious, to get a third seat in Quebec, the NDP's closest bet is Jeanne-Le Ber, in which they're 6.6 points behind the Bloc. They'd need a huge uptick.

  9. William,

    I don't know what you mean.


    I'm going to be writing about the NDP in Quebec in this afternoon's post, so I'll respond to most of your points in it.

  10. After talking with my relations, and reading one of the political stories in the Globe, I wonder if people are putting too much faith in the data and confusing its complexity for completeness.

    I mean, when a writer at the Globe looks at a poll that has the CPC at 39%, the LPC at 29%, or some commenters here look at a projection of 144-150 and think that the majority is out of reach; I think they've misunderstood that there is variance possible in these projections as well as the polls; and that it doesn't much of one.

  11. @William
    Westmount-Ville Marie, QC also looks interesting, with a 10% of the libs (down from around 14% a couple of weeks ago), but no danger of splitting the left.

  12. In the graphic at the top of the page, you've got the Bloc in the bottom left of the Quebec popular vote projection (which is good, as it makes the regional votes look more consistant), but the two big pies for national vote and seat projection have the Bloc slice at the top. Perhaps move it between the Conservatives and Liberals?

  13. I remember the 2006 election when the Tories suddenly surged in Quebec and all the commentators were scratching their heads saying "OK, Bernier will win in Beauce and maybe Cannon in Pontiac and maybe Verner in Louis St. laurent...but after that where do they get their fourth seat?" as we know, they got 10 seats on 24% of the vote!

  14. hosertohoosier18 April, 2011 10:59

    Can we see a projection with just the Angus-Reid poll? I'd be curious to see what an NDP-Liberal tie would look like.

  15. I think I prefer the current make-up, as it puts the Liberals and Conservatives next to each other.

  16. Terrific site - thank you for your daily feed of analysis Eric.

  17. Hosertohoosier,

    Wait a few more days and maybe you'll find out!

    I have something else in mind today, so maybe if there is a future poll with a Liberal/NDP tie I can do that.

  18. And thanks, everyone, for your nice comments.

    The internet is full of people trying to tear others down, so it is truly appreciated when people take the time to say thanks. I'm not sure I deserve it, I've honestly been humbled by the (modest) amount of success this site has had during the election, and I'm very happy that people enjoy the stuff I produce.

    And a thanks to all of you for your helpful and informative comments - I've received emails from across the country with little tidbits of valuable local perspective, and I have learned a lot. So thanks.

  19. Thanks very much for your site, Eric. Although I don't usually muddy up the comment threads, I do visit this site at least once a day, and usually two or three times a day.

    It's, without a doubt, the best Canadian polling resource on the Internet. Keep up the good work.

  20. Eric:
    The Nanos poll on CTV's website has the CPC up to 39.8% in Canada and up to 45.2% in Ontario. The 39.8% is over 1% higher than you are pegging them at nationally. The 45.2% in Ontario up from 42.8% is a huge 1 day gain especially on a 3 day rolling average. Nanos puts the Conservatives ahead of the Liberals by almost 9% in Ontario now.
    Why all the jumping up and down about what appears to be bad news for the CPC and no mention of the results from what is usually accepted as the top pollster in Canada.
    I do respect your work very much, but it seems that sometimes the reporting is almost gleeful when it hurts the Conservatives and very slow and sullen when it is good news for them. Are polls and pollsters not supposed to be very neutral? I am disappointed.

  21. DL, I think the difference between 2011 and 2006 in Quebec is that in 2006 it was two big shifts - Conservatives up and Liberals down. In 2011, there is a big increase for the NDP, but the decreases of the other parties are not as large.

  22. Anonymous 11:16,

    Perhaps it is my own hope that the race will be interesting is what you are seeing. If the Conservatives were roaring ahead, I'd be trumpeting that because that would be an interesting turn of events.

    But the projection is the projection - those Nanos numbers from today were included. If they hadn't been, we would have seen the Conservatives dropping more. I'm only discussing in this post about how the projection has changed.

    I'll post about the individual polls this afternoon, so I will certainly be making note of the changes that we've seen in the newest Nanos. But Léger and Angus-Reid have the Tories dropping, so we must take that into account as well. One poll is one poll, after all.

  23. What we are seeing is very stable polls, with very small fluctuations moving a lot of seats one way or another, which is exactly what I said would happen. We are in either a strong Tory majority or a very weak Tory minority within the polls margins of error.

    Personally I believe the hockey games are depressing Tory numbers,aka yesterday in BC.

    Also the polls are still showing the Tories 4% above their final polls in 2008. There's no guarantee the polls are off in the same manner this time, but it's a good possibility.

    Either way we won't know until election day if the numbers stay on the cusp, which they have the entire campaign.

  24. Hi Eric,

    I have a question. Do the polls (and by extension, your prediction model) assume something about the demographics of people who actually vote? Specifically, do they put less weight to younger people voting intentions, given that very few of them actually vote? This may become important if thanks to social media, vote mobs etc. young people will show up in large numbers to vote on May 2nd (the "Obama effect").


  25. I don't know if pollsters weigh the youth vote by their proportion of the general population or by their turnout rate. I'll ask around, since a lot of people have been asking me this question.

  26. I agree with Simon. It would be a statistical anomaly for the NDP gains in Quebec to be uniform. There is going to be dispersion around the gains. In some ridings the vote will be 40% +, while in other ridings it will be 10% or less. While in most ridings it will center around the proportions shown in polls. I think this means there are a dozen or so ridings up for grabs for the NDP in Quebec.

  27. @tyler:

    Or it may mean no new seats for NDP - depending on where the "NDP surge" is more pronounced. If it's more pronounced where they still have no chance of winning, or where they will split a vote and elect a Conservative instead of incumbent Bloc or Liberal, then this localized "surge" will be wasted.

    So it could go both ways, and we can only speculate at this point.

  28. Anonymous 11:25 - Don't forget the provincial NDP leadership convention was yesterday, as well. By your logic - and I'm not (necessarily) saying that it's wrong - that might have depressed the NDP support, as well, and in BC, as with much of Western Canada, they draw directly from each other.

    However, there're also effects like weather (does good weather mean that more families are outdoors?) and even season finales of popular TV shows.

    However, it's probably worth pointing out that some of the same effects could well be in play on May 2nd. Let's say that the Canucks are playing in an Eastern or Central time zone on that day - it would mean that folks coming home from work might reasonably not stop to vote, because the puck would be dropping at 5 or 6 pm.

    So all we can say (like always) is to not put too much stock in polls, but that large movements are always interesting, no matter what confounding factors we might be able to hypothesise. Hence Eric's earlier comment - it's not so much about the Conservatives going up or down, it's more that he (and many of us) are hoping this election becomes *interesting*!

  29. Yes, yes, youth vote question is king. Good reason to think they will turn out in greater numbers, perhaps much greater number (though I'm open to arguments that they will turn out to be more smoke than fire). This is exactly the kind of thing that could throw pollsters way, way off course. A post on how the major firms are weighing youth vote would be beyond lovely. Doug Johnson Hatlem

  30. Westmount keeps eluding the NDP. I'm guessing a large chunk of their support is from McGill and Concordia students who very well may be out of town when it comes time to vote.

  31. I've gotten responses from some major pollsters, all saying they weigh young Canadians according to their proportion in the general population, not according to their historical turnout rate.

    So the polls could potentially not be reaching enough young Canadians, but in terms of how they weigh the responses they do get turnout should not be a huge factor.

  32. I do find the results odd vs what I have been hearing. Locally a few members of the Conservatives have told me that the Greens (of whom I am a riding president thus have a bias POV) will do the best ever in our riding based on what they have been seeing at the door. More sign orders (climbing quickly - hopefully we have enough) more interest. To be honest it caught me off guard a bit lately as the Nanos polls keep us sub 5%, the others are not growing, May was kept out of the debate thus pushing us down further. There is something odd this election - I wonder if the polling methods are getting less accurate (cell phones and the like) or if there is a change in voting patterns or if it will all amount to nothing and we end up as listed here.

    May 2nd will be very interesting. If it is a tight race and Greens do well then EKOS looks good. If the Conservatives win by a mile and the Greens disappear then Nanos looks good. If the Liberals actually win or some other unexpected final result occurs then the pollsters look really, really bad unless they catch it in the last week.

  33. Comment - Part 1:


    All polls in Canada are stratified by age, so that the percentage % of folks in any given age group (ie 18-25) matches the population as a whole. - The do NOT weight to actual past turnout.

    How this works in practice, is the lower turnout among youth shows up as there being substantially more "undecided" voters among the younger age group, so when the undeided are excluded from the "horserace" question, they will present in smaller numbers that their portion of the actual population

    Please note this stratification applies to the totality of the NATIONAL sample, NOT individual subsets. The consequence of this is that the age balance in any individual subset can be (and often is) massivey wrong. For example in any given night the national % of young people will be right, but there could be way too many young people in say Ontario and too few in Quebec...

    This is why we see such huge (and irrational, and false) regional shifts in the polling while the national topline stays so constant...

    A good way to think of this is if I randomly flip 1000 coins... I will get for the whole sample "about" 500 heads and "about" 500 tails.. plus or minus some statistical noise...

    If I take a daily subsample of 30 coins and call it "Atlantic Canada" some days I get 12 heads, and the next day I get 18 heads... The likelyhood of the "Atlantic Canada" coin being heads or tails (obviously)has not changed, but because the subsample is so small, there is a huge amount of artificial "volatility"....

    The Vorlon

  34. Comment - Part 2:

    The use of screening questions (aka the EKOS Voter Intensity Index) is a whole can of worms Canadian Pollster avoid.

    Given that only 59% of Canadians voted in 2008 it makes obvious logical sense to try to sift your sample down to the "likely voters" - but the mechanisms to do so are far less developed in Canada than they are in the US.

    This "likely vote" issue "could" have substantial impacts in 2011 as turnout seems headed even lower than in 2008.....

    In 2008 the polls, as a group, under-projected the Tories....

    The actual margin of Tory victory was about 11.4% better than the Grits...

    By contrast...

    EKOS projected a 9 point Tory margin

    NANOS projected a 7 point Tory margin

    Harris Decima projected a 9 point Tory margin

    Angus Reid projected a 10% margin

    Strategic Council (now kinda sorta Harris/Decima) projected a 5 point Tory margin

    The average of these 5 polls was a Tory margin of 8%, versus the actual result of 11.4% - an average bias of 3.4%

    Flash forward to 2011....(as of April 18, 2011)

    Tories Grits Dippers Bloq
    NANOS 39.8 29.8 17.4 8.6
    A/R 36 25 25 9
    Leger 38 26 22 8
    EKOS 35.3 27.8 18 9.6
    IVR 39 28 17 9
    Average 37.62 27.32 19.88 8.84

    The average of the 5 most recent polls is a Tory lead of 10.4%

    This is about 1% less than the actual result in 2008, but 2.4% better than the average of the pre-election polls of 2008.

    If you believe the polls, the result in 2011 we be a whole lot like 2008.

    If you think the polls will show the same structural bias as they did in 200 and the same "add 3.4% to the Tories" average applies, the a Tory margin of 10.4% + 3.4% => 13.8% puts the in semi-solid majority territory.

    The big unknown here is voter turnout.

    In 2008 turnout was about 59% (a historic low) and polling suggests in may be worse in 2011.

    All the polls suggest that the Tory vote is more motivated.

    The EKOS poll (the least favorable current poll to the Tories) shows the Tories have a substantial lead on their "Voter Commitment Index" which measures likelyhood of voting. - The Angus Reid "Absolutely Certain to Vote" subset shows similar outcomes.

    Now if the 3.4% gap between the pollsters and the actual result was a fluke in 2008, or the inevitable product of using all voters versus "likely voters" remains to be seen, but if you are a Tory, and you believe the 10.4% + 3.4% => 13.8% formula is valid, you have to be thinking "majority government"

    The Vorlon

  35. @Anon (11:16)

    I wouldn't say Nanos is usually accepted as the top pollster in Canada. Usually considered to be among the top, perhaps. Most active among the top pollsters in this election, definitely (and thus the most talked about). But I wouldn't say they've got a clear cut better reputation than all (or necessarily any) of the other big pollsters. Certainly not enough of one to be generally accepted as *the* top pollster.

  36. @The Vorlon

    Regarding stratification being national and not regional or some other subset, that is not true.

    I know for a fact that some pollsters do fully nest their demographic quotas and weighting stratifications by age, gender and province/region. So the % of responses they get from 18-25 males in Ontario is in proportion with the % of 18-25 Ontario males in the general population.

    I don't know that all pollsters do this, and those that do may switch it up from time to time (especially with one-off surveys).

  37. In some ways NANOS "benefits" (if that is the word) from the Rasmussen Effect. In the US Rasmussen reports generates so many polls that they, by sheer volume, they to shape the debate and dominate the discussion.

    NANOS does a poll every day, we talk about them everyday...

    The others do a poll every 1 or 2 weeks... we talk about them less...

    Regarding Canada, NANOS was absolutely dead on in 2006 (as in 1 in a million longshot accurate) in 2008 they were off by a fair bit....

    In 2008 they said the Tories were up by 7% - the actual value was 11.4%..

    Given the real world complexities of polling, an error of 4.4% is hardly a disgrace, but it's not Svengali like accurate either...

    NANOS deserve to be in the mix, no more, or no less than any other poll....

    "Top Pollster" is a bit of a stretch in my estimation....

  38. Hi BT,

    Let me extend my remarks.... the online/internet pollsters DO indeed as you suggest stratify by age/gender/geography within the regional subsets.

    My comment should have been restricted to live operator telephone based pollsters...

    Given the (wretched) experience with online surveys in the US election process, I tend not to think of these onine panels as "polls" actually.. Though In Canada they do seem to do pretty well historically...

  39. Eric,

    Let me add my voice to the cheers for your site and your work. My "look forward too most" web surf of the day.

    I don't know if this would be possible but I would really be interested in your take of what sort of swings we would need to see (and where we would need to see them) to answer certain "what ifs" that are now being talked about in the MSM: a Liberal collapse to a surging NDP leading to a Tory majority? an NDP official opposition? a bleed of soft-nationalist Bloc support to the NDP in Quebec helps elect whom?

    Don't know if the interactions are too complicated to parse out but given the relative stability of the numbers, this is shaping up to be a very interesting election.

    Oh, @Anonymous 11:16 I think Eric is falling into the trap of any good journalist, indeed any good writer: you "root" for the story.

  40. Thanks Highlander. I'll be doing the NDP's ceilings in this weekend's ceiling post, and will be looking at the NDP in Quebec this afternoon. Perhaps those will answer most of your questions.

  41. With all the discussion of the difference between the polls last election and the actual result, one point worth remembering is that there was an upswing in the economy the day of the last election which I suspect results in an upswing of support for the tories. i.e its possible the poll were accurate, but of course could not predict the last minute shift.

  42. @The Vorlon

    I was also referring to interviewer-conducted phone surveying.

    My experience seems to be the exact opposite of yours - interviewer conducted phone surveys are more likely to have fully stratified demographic quotas, while self-conducted surveys (online, IVR) are more likely to have non-stratified demographic quotas.

  43. 2 few points worth considering:

    -- Jack Layton has a strong post-debate wind in his back
    Approval Rating (from the last Angus Reid):
    -------------Layton 50%,
    -------------Harper 33%,
    ------------Ignatieff 24%,

    Iggy is ranking third amongst the 3 leaders !!

    -- BUT the NDP leaners are very prone to switch their vote -- for anti-tory majority tactical voting reason, or because they believe that their vote would be wasted.

    So IF Layton's profile keep on rising and Iggy's keep on declining, and IF Layton and the media keep on hammering that the NDP can achieve SECOND rank, behind the Tories, and keep on attacking the Red Team, the soft NDP votes might solidify.

    Exit question: What would be the effect of that (the marginalisation of Iggy's profile and the solidification of the soft NDP's voters) on the prospect of a Conservative Majority?

  44. One last thing:

    On the *Vote Retention* front, the Tories score higher:

    84% of those who voted Tory in 2008 would vote Tory again.

    For the other parties, its:
    Liberal: 70%
    NDP: 78%

  45. Hi BT,

    Maybe they just do things different in Canada than the US, but I am not sure how it is even possiblem give sample sizes, for a Canadian Regional subsample to properly stratified for age/gender/education/geography/race/income etc..

    In the US the semi-standard weighting standard to to "quota call" till you have filled your quota in the 48 "cells" the US Census bureau uses.

    I suspect the Canadian census data will have a similar number.. you simple cannot balance all of that in the 100 person "Atlantic Canada" sample... Nationally yes, Regionally no....

    I would offer the following common sense argument as well...

    NANOS has be pretty darn steady at a 10% National Tory lead +/- a bit of statistical noise.

    Yes the regional sample have been nuts...

    The Tories have been upp 12 and down i8 in Ontario..

    The Grits have been down 18 and up 9 in Atlantic Canada...

    The Dippers have been below 10% and above 20% in Ontario...

    Despite these HUGE regional swings.. the topline national number has been a rock steaddy 10%...

    What seems more likely to you.... That all these wild region swings were real and by some micacle managed to all cancel eachother out at the national level to keep things at 10 Tory lead..

    Or was it that the regional samples are often wildly unbalanced as to age/race/gender/income/education? and hence the volatility is an artifact of unbalanced sampling and not reflective of actual changes in the electorate....

    The Vorlon

  46. @Katchecon

    More of the same re Iggy crashing and burning....

    Harper 95.9
    Layton 58.1
    Iggy 40.5

    "Jake Layton - The Leader of Her Majesty's loyal opposition...."

    Has kinda a nice ring actually....


  47. Regarding a possible Quebec breakthrough for the NDP like the one the Conservatives achieved in 2006 I think it would consist of these seats. Not necessarily in order other than the first two. Dont think they will actually win all these ridings, but hopefully they win most of them :).

    Outremont: (They have it already)
    Gatineau: (3 points behind the Bloc, with all three opponents down in Quebec)

    Abitbi-Baie James-Nunavik-Eeyou: Strong local candidate, BQ incumbent only had like 38% last time
    Hull-Aylmer: Liberal forever, literally, but NDP was a solid 3rd last time and not too far behind.
    Pontiac: Yes Cannon has barely 32% of the vote and the NDP are strong in some portions of this riding at about 16%. Vote splitting.
    Jeanne-Le-Ber: Liberal/Bloc splitting at around 33% NDP solidly in 3rd, one of the poorer/working class ridings in the area.
    Laval-Les-Iles: Longtime Liberal incumbent retiring, strongly federalist riding, may be more up for grabs than we think.
    Laval: NDP in fourth but a fairly split up vote here may lead them to victory if they are in the high 20s.
    Alfred Pellan: Another riding with split federalist votes and a weaker BQ MP.
    Westmount Ville Marie: NDP targetted it last time obviously, they are solidly in second and would be primed to take advantage of any Liberal vote loss.
    Beauport Limoliou: Conservative incumbent here but it is very close between the Conservatives and BQ and with both parties down, a 30% share may take this one if the NDP keep moving up.
    Lac-St Louis: Perceived to be a 2-way race between the CPC and Liberals, the NDP have a solid base here and the Greens have a lot of votes as well that would likely go NDP in the absence of a higher profile national campaign.
    Gaspesie Iles-de-la-Madeleine: Heck they own a neighobring riding in NB, and people remember that the CPC led this riding for quite a while in 2006 during their breakthrough.

    Also interesting to look at a number of ridings where the NDP are second to the BQ in suburban regions just outside of Montreal for future elections.

  48. Doug here. First, my congratulations to you for running an excellent site. This is very informative, and I appreciate very much that you have kept it factual and impartial.

    I, too, would be very interested in any information or links which break down voters' intentions along age demographics. There was a recent graph in the Globe and Mail which illustrated a very different parliament if the only voters were in the 18 - 24 year old demographic. I'd be very interested to see the 65+ demographic.

    My best wishes to you,


  49. Thanks Doug.

    That graph in the G&M was part of an article that I had written, the end of which did look at only older voters. The result was a massive Conservative majority, with little Bloc representation and no NDP MPs.

  50. @The Vorlon

    Nested for Age-Gender-Province/Region only, and sometimes language within Quebec (though just as likely not nested with the rest). They generally won't nest income, education or urban-rural split if they are being used unless one or more of the above is non-nested, and race is rarely used to stratify in Canada compared to the US (unless it is notable in regards to the topic of the survey). You're really looking at 6 to 10 quota cells (3-5 age groupings) per Province or region. That is how it's done.

    My experience has shown some significant differences in how US surveys operate compared to Canadian surveys. It's actually been an issue I've dealt with before for cross-border work when assumptions are made. One was how national representation was handled.

    Nanos, running daily polling during the election, is not likely doing full nesting of age/gender/region. Too tight of a time frame to reliably fill those difficult quotas. The various companies running weekly trackers are more likely to be doing so. The big tip-off for Nanos is the shifting regional MoE indicating soft quotas instead of hard.


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.