Thursday, May 6, 2010

A Tale of Two Polls - 4 or 7 pt CPC Lead?

Two polls were released this morning, one from Nanos and one from EKOS. Lately, these two pollsters couldn't be any different. EKOS finds rock-bottom support for the Conservatives and Liberals, while Nanos usually has them both higher than everyone else. According to Nanos, 70.4% of Canadians support one of the two big parties. For EKOS, that number is 59.2%.

We'll start with Nanos, who last polled in early March.Compared to that poll, at 37.2% the Conservatives have gained 2.5 points. The Liberals have dropped 1.4 points to 33.2% and the New Democrats are down 1.6 points to 16.2%. The Greens are down 1.4% to 3.8 points.

It's difficult to reconcile these numbers with other polls - which I know can be a fool's errand. But no one has had the Liberals over 30% in a very long time, and the Conservative result will be one of the highest in my model. And the Greens below 4%? I suppose it could be true, but I doubt they will get 0.3% support in Quebec and 0 votes in Atlantic Canada.

Anyway, in Ontario the Liberals drop 0.1 points to 41.6%, while the Conservatives are down 2.1 points to 37.1%. The NDP is up 2.1 to 16.7%.

In Quebec, the Bloc Québécois jumps 6.4 points to 37.9%, while the Liberals are down 4.3 to 26.7%. The Conservatives are up 1.7 points to 23.5%, one of their best results ever. The NDP is down 0.1 points to 11.6%.

In British Columbia, the Conservatives have lost 0.7 points but still lead with 36.1%. The Liberals are up 4 points to 32.2% and the NDP is up 2.5 to 25.6%.

In the "Prairies", which Nanos defines as Alberta, Saskatchewan, and Manitoba, the Conservatives are up 7.2 points to 54.7%. The Liberals are down 4.4 to 24.6% and the NDP is down six to 13.8%. In Atlantic Canada, the Liberals lead with 43.7%, followed by the Conservatives (up 13.4 to 38.4%) and the NDP (down 11.6 to 17.9%).

Using this site's current projection for Alberta and the Prairies, the Conservatives win 66 seats in the West, 41 in Ontario, 10 in Quebec, and 9 in Atlantic Canada for a total of 126.

The Liberals win 19 seats in the West, 55 in Ontario, 16 in Quebec, and 22 in Atlantic Canada for a total of 112. A good result for them.

The Bloc wins 49 seats.

The NDP wins 10 seats in the West, 10 in Ontario, and 1 in Atlantic Canada for a total of 21.

Now, EKOS.
Compared to last week's poll, the Conservatives are up 1.2 points to 33.1%. The Liberals drop again, 0.5 points to 26.1%. The NDP is down 1.6 to 16.0%.

Coupled with Nanos, this indicates the little bump the NDP had may have disappeared.

The Greens are up 0.6 points to 11.5% and "Other" is down 0.2 to 3.1%.

In Ontario, the Conservatives are steady with 36.3%. The Liberals drop a point to 32.5% and the NDP is down two to 15.7%. The race is close in Toronto, with the Conservatives at 37.8%, two points ahead of the Liberals. In Ottawa, it isn't as close, with the Conservatives leading with 40.6%.

In Quebec, the Bloc is up one point to 40.4%. The Liberals are up two to 20.9%, as are the Conservatives at 18.4%. The NDP is down three to 9.6%. In Montreal, the Bloc leads with 36.4%.

In British Columbia, the Conservatives are up two points to 29.6%, followed closely by the NDP at 25.5% (up three) and the Liberals at 25.1% (up two). In Vancouver, the Conservatives have a narrow lead over the NDP, 32.7% to 31.2%.

Elsewhere, the Conservatives lead in Atlantic Canada with 38.6%, where the NDP has dropped seven points. The Tories also lead in Alberta with 49.8%, down five. In the Prairies, they lead with 48.4%, up 10, while the Liberals are down eight to 19.8%.

The Conservatives win 66 seats in the West, 50 in Ontario, 8 in Quebec, and 11 in Atlantic Canada for a total of 135.

The Liberals win 16 seats in the West, 41 in Ontario, 14 in Quebec, and 17 in Atlantic Canada for a total of 88.

The Bloc wins 53 seats in Quebec.

The NDP wins 13 seats in the West, 15 in Ontario, and 4 in Atlantic Canada for a total of 32.

So, two very different polls. The big culprit seems to be the Greens, as I think we can all agree it is unlikely the Greens would drop to 3.8%, losing almost half their support, or double their support to reach 11.5%. So, they are more likely somewhere in between, meaning there are four points to distribute elsewhere. That changes both polls quite a bit.

More importantly, though, is where the polls are similar. They both show the Conservatives making modest gains, and the Liberals losing ground. They both show the NDP down and the Bloc up. They both show a close race in Ontario and British Columbia. These are the things to take from these two polls.


  1. I've felt for a while that the Green results have in fact been a mixture of actual Green support and a "none of the above" support.

    Nanos may be closer to the actual Green support.

    At best we wait for further evidence of either.

  2. I don't think its at all unreasonable that the Greens would end up with about 4% of the vote. That's what they got in the the 2004 and 2006 elections. In 2008 their vote was artificially high since everyone was talking about the environment, there was all this saturation publicity about Elizabeth may being the "new fad", there was all this publicity about her being in the debate for the first time and her non-aggression pact with Dion etc...but all that has since collapsed. There has been ZERO coverage of the Green party in the last few months. The media now sees May as over-rated and yesterday's news. I expect them to get about 4% of the vote next election.

  3. John,

    I just have a question for you.

    What do you make of the difference in the level of support for the greens in the two polls.

    I am not saying either is right or wrong, but I would just like your analysis on the two results.


  4. I expect them to get about 4% of the vote next election.

    Certainly fits with my impressions. Nanos seems more accurate on this.

  5. Let's also not forget that we had four federal byelections last November and on average Green support was THREE PERCENT. So the nanos poll seems par for the course.

  6. Most likely the Greens won't be apart of the debate next election, therefore they probably won't have as high of a profile they had in 2008 which will cause there support to drop.

  7. By-elections are horrible indicators for national trends, especially for third-tier parties. They are only useful in gauging support for the government and the main opposition.

  8. I don't think its at all unreasonable that the Greens would end up with about 4% of the vote. That's what they got in the the 2004 and 2006 elections.

    It's also not unreasonable to think that support for third parties is higher since 2008, since the economy isn't where people think it should be, and since both of the big two parties have been having problems. While the Greens may have trouble getting above 10% when a vote actually comes (due to increased scrutiny, lack of organization and so on), it's not crazy to think they have more support (even if weak, "none of the above" type support) than they have had in the past.

  9. Nanos 37/33/16
    Ekos 33/26/16

    Normalized using Eric's ratings pollster vs pollster:

    Nanos 38.5/30.5/16 = 8pt tory lead
    Ekos 34/25/17 = 9pt tory lead

    So it would seem that the 2 pollsters agree with the results (as far as the in house differences between them historically go)

    Eric rates Ekos as a little more accurate than Nanos.

    So (very) crudely.... 36/27.5/17

    37.5/26/18 - 2008 election
    36/27.5/17 - 2 poll crude average

    Doesn't look that different to me. Couple seats change hands??

  10. Other ekos notes:

    BC looks like a nice 3 way race. and with gre3ens riding high at 15% there... could see a split and a seat.

    Why are the Greens running second in Alberta?? Don't think so.

    Looks like the liberal surge in sask/man was the 1/20. Pols is back where it was a week ago.

    Liberals in trouble in Ontario again... tories back to a 4 pt lead. What is the area that Ekos defines as Toronto??? It must be bigger than the 416 if the tories are leading there.

    Bloc above 40. I don't think I have seen that for awhile.

    NDP below 10 in Quebec?? That doesn't seem to agree with other recent polls some of which put them over 20.

    Atlantic Canada seems to have experienced the same as the Prairies.... NDP back up almost 8, Greens back down.

  11. We've all been looking at these polls for several months. What I think I'm seeing is:

    No Change !

    Basically everybody is around the situation after the 08 election. There is NO significant shift anywhere.

    So all that can be said at this point is that NOTHING is Selling!!

    In other words nobody is getting their message across to the public in a meaningful way. No policy is resonating, no new program is grabbing attention. Stagnation really.

  12. The 4% result for the Greens is fairly unlikely as 0% support east of Ontario seems extremely unlikely. In the by-elections a lot of GPC members were wondering why so little effort was put in, but the general feeling was that we had little chance of an upset, or even a strong showing given the resources we knew the big 3 were putting in, thus saving cash for a future election was important. Not to mention the memory of 2008 when we put in a ton of effort and resources to winning in Guelph (and appeared to have a real shot) just to have the by-election cancelled for a 'real' election, killing our momentum.

    Basically when the next election comes the Green Party is likely to either A) gain votes and win a seat or two or B) collapse and disappear. The media would probably prefer option B and to see the NDP merge with the Liberals and have the BQ die off (much easier to report on 2 parties than 5). The average voter though seems to like having options given most polls are showing CPC/Liberal support at all-time lows (if you merge Reform with PC in the 90's).

    Would it be possible to list what the specific question is that Nanos and EKOS use in their polls? The wording could explain part of the odd result. Also age by region could explain a lot (seniors are more Liberal/CPC than the sub-65 crowd).

  13. This comment has been removed by the author.

  14. 20% for the Greens and Others combined in BC? More likely 10% with 10% apportioned to the other parties.

  15. Here are these 2 polls adjusted to the bias/error shown in the 3 polls by these companies the last 4 days before the 2008 election. You know the one, where the pollsters actually get to compare their numbers to the real election numbers not to other pollsters as Eric provides his adjustment factors.


    CPC 33.1 + 3.3 = 36.4
    Lib 26.1 + .1 = 26.2
    NDP 16.0 - .6 = 15.4
    BQ 10.2 + .1 = 10.3
    Grn 11.5 – 3.7 = 7.8


    CPC 37.2 + 4.5 = 41.7
    Lib 33.2 – 1.0 = 32.2
    NDP 16.2 – 3.6 = 12.6
    BQ 9.6 + .2 = 9.8
    Grn 3.8 – 1.6 = 2.2

    Nanos poll puts the CPC in Majority territory and has the core opposition support going to the Liberals (by default? Or a focusing the ABC forces?)

    The NDP is being punished for having Pat Martin out in front of the Jaffer posse and looking mean and intellectually challenged.

    The Green support is totally disappearing. They tied their wagon to AGW and that is so yesterday. There are absolutely no awards being given to Al Gore and David Suzuki for the next 5 years. Last election the Greens was riding the all time glory of the Al Gore religion. They can come back if they could focus on clean water, real air pollution... the kind that is mainly limited to big cities and having more and better parks. This won’t happen with Ms. May.

  16. 49 Steps: What do you make of the difference in the level of support for the greens in the two polls.

    A bit of a head-scratcher, isn't it?

    The interesting point isn't the comparison of EKOS's 11.5% Green support (which I'd adjust to 11.3% after fairing the curve) versus Nanos's 3.8% (which I'd put at 4.4% or thereabouts). The puzzling inconsistency is the way that EKOS numbers have shown a slow but persistent climb in Green support since the last election, while Nanos numbers have shown a decline over the same period. My hairy eyeball shows the same effect in the NDP numbers at a much-reduced level. (Quantitative refutations are welcome.)

    Why the difference, and whom do we believe? Most importantly, what will happen in the polling booth?

    The most obvious difference between Nik Nanos and Frank Graves is the presence or absence of party prompting. The Green Party had a significant mind share after the last election. Since then, party advertising has been minimal and earned media opportunities have been few and far between. The party doesn't pop into people's minds when Nik calls and they have to answer right now.

    By contrast, the larger parties have a daily muckfest in Question Period, backed up by photo ops and running gong shows. You may or may not like them but it's hard to forget them.

    The Dipper numbers are consistent with this hypothesis. The effect is subtle because Jack Layton and the NDP earn considerable air time--but not as much as the two largest parties.

    So how does this play out on the hustings?

    When the writ is dropped, Green visibility will climb. Our riding will once again be blanketed with Green signs before the other candidates have pounded their first stake. The nightly reports will always include a Green update because it's hard to blow off a party that has a third as many supporters as the first-place party. The electorate will be constantly reminded that Greens are a viable alternative. And when voters enter the polling booth, the piece of paper in their hands will say "GREEN".

    To me, that's closer to the EKOS polling context than the Nanos context. I therefore attach more significance to the EKOS trendline. Add your own views on house effects and GOTV, plus the subtle difference between "prefers" and "will actually go to the polls and vote for".

  17. Have you considered soming up with a correction macro for the polls that lump Alberta and the prairies together?

    For example, take other polls that do separate them, calculate new combined region totals for those, and then use the Alberta and prairie totals' differences from that new average as the correction for Nanos polls.

    So, for example, if the other polls together tend to place CPC support in Alberta 7 points above the average of the two regions, with the prairies support 9 points below, adjust the Nanos numbers in that way to produce your songle-poll projections.

  18. Interesting results in both polls.

    What did I tell everyone about the NDP results, eh? What did I say - it'll last a week or so. Never doubt the Volkov.

  19. "Would it be possible to list what the specific question is that Nanos and EKOS use in their polls?"

    It is common knowledge that EKOS says "For Liberal press 1, for Conservative press 2, for NDP press 3, for Green press 4 and for some other party press 5" (this has the effect of psychologically juxtaposing Green and "other" with the real parties. Nanos just asks something along the lines of "Which party would you vote for?" Period - no reading of names. The respondent has to volunteer "I would vote Liberal" or "I would vote NDP" etc...needless to say - Green support seems to evaporate when people have to say "Green" without prompting - which tells us something about how soft their support is.

  20. Interesting. Ekos says they managed to survey both those people with landlines and those people with cell phones. The usually story is that young people are less likely to have landlines and more likely to vote greens (or, at least, say they will vote green, which I suppose isn't the same thing), and so Green support is understated in traditional polls. That methodology might explain the surprisingly high Green support in the Ekos poll.

  21. That's an interesting idea, Ira, but I'm not really sure it would be worth the effort. I think Nanos is the only one that doesn't break down the Prairies into AB and SK/MB. And, in any case, those two regions are the least at stake.

  22. Shadow,

    Sorry, I'm not sure I understand the distinction between "committed" and "decided" voters. I'd have thought that that was just a difference in terminoloy between Nanos and Ekos. Indeed, I note that in the Nanos poll the respondents are broken down into two broad catagories "committed" voters (who are then broken down by party preference) and "undecided" voters, which suggests that Nanos uses "committed" voters in the same way that Ekos uses "decided" voters.

    It is interesting that Nanos has a higher level of undecided voters than Ekos (22$ vs 14%) which may reflect differences in their methodologies (Nanos, as I understand, does not probe for parties, maybe Ekos does.)

  23. Now, a more interesting question is how Nanos gets Green support in Quebec at 0.03 percent with only 198 respondents. Even if only one person responded, you'd have thought that would be slightly more than 0.05%. Either they're weighting their respondents to ensure that the sample composition reflects that of Canada (which is not uncommon, but which isn't disclosed) or someone made a booboo with the calculator.

  24. Interesting to think about the methodology of the two pollsters and how it compares to election day.

    When you vote you do see all major parties (Greens qualify as they run candidates in as many ridings as the big 3) plus a few oddballs depending on your riding. So asking people from a list makes a lot of sense as that is what they see, you don't have to remember your local candidates name or party when you go into the booth, you just have to find the name you like and mark an X next to it.

    However, how much adjusting each does for likely turnout is another issue. We all know the older you get the higher the odds of voting (until you cannot walk to the booth anymore). We all know the sub-25 crowd would prefer pretty much anything to getting out and voting. We also know that if a real controversy erupts that we will see a significant uptick in voter turnout but otherwise will see it continue to go down thus rewarding parties that are good at GOTV.

    Are there good numbers for Canadian participation in the vote by age group/gender/region out there? It is easy to find it for the USA but not so much for Canada. It would be interesting to take the polling results (when they provide an age breakdown) and see what happens if you adjust results to match likely voter turnout rather than general population.

  25. This dissection is all very interesting but I see no attempt to quantify the other grouping.

    "No intention of voting" !

    Last election that was a significant number. So all we are seeing right now is the voter who intends to vote. Yet last time I think we nearly had 50% of the electorate not voting? Does that tell us anything ??

    Eric do you have any figures ?? Thanks

  26. I don't like the idea of pollsters listing the parties to people. Not listing the parties only seems to really effect the Greens which says alot about their support, if you don't no if you're going to support them without being promted then I doubt many will support the Greens in the end.

    I don't have a problem with the Greens, though I'd rather see them do well after the Bloc dies off, but I think if a pollster is going to list parties they should either just list those with sitting MP's or list every party recognized by Elections Canada. I'm willing to bet if a pollster started listing the Progessive Canadian Party or the Pirate Party as options then their support would also rise especially over a year and a half.

  27. An interesting thing I just checked. If you take regional results for the two polls and figure out total responses you can see that Nanos is dead on - ie: no regional adjustments outside of number of people asked - while EKOS is off by a fair margin - i.e. regional adjustments made after getting the results.

    So then I took their regional figures, adjusted by number of seats per region (since seats are all that matters) and looked for any changes.

    For EKOS no party shifted by more than 0.4% (CPC goes up by 0.4%, Liberals stable, NDP up by 0.3%, Green & BQ both down by 0.3%). For Nanos there was one big shift - the Liberals go up by 0.7% to 33.9%. Nanos was significantly weak in their weighting for Ontario & Atlantic Canada (11-12% less strength than they actually have) while over representing the prairies (11%) and BC (17%).

    Doing this screamed to me that Nanos is not doing a strong check on regional spreads before merging into their national figures and makes me wonder if they double check by age/gender to the degree they should (women being less likely to vote CPC than men, young more likely to vote NDP/Green than older).

    Of course, no change is equal to the margin of error nationally but when we see two polls showing big differences one has to wonder why and from what I can tell Nanos was sloppier than EKOS in their methodology.

  28. Looks to be like the difference is weather or not the Greens were prompted.

  29. This comment has been removed by the author.

  30. There's much to be happy with in the Green tea leaves from EKOS this week.

    The national Green number of 11.5% is a shade high based on eyeballed curve-fairing, so I'd put the "real" value at 11.3%. This continues the slow-and-steady upward trend.

    The BC level of 15.3% is very encouraging when the other parties are so evenly matched. Greens will win first where other parties split the remaining vote, and BC is ripe for that outcome.

    Alberta again provides some entertainment with Greens in second place at 16.6%, better than the number to the west. However, the Tories still have half the vote so Green seats will be much harder to win there. In any case, this is comedy relief, not reality. The Grits are still ahead of the Greens in Wild Rose country--for now. The crossover is within sight, though.

    Saskitoba, Quebec and Atlantic Canada all show reduced Green support relative to recent polls. Were those polls too high, are these numbers too low or has there been a real change? As long as the national Green levels continue to climb, real or apparent fluctuations in those three regions are not a cause for worry. As I've commented in past, the curse of the Green Party has been its national appeal in a first-past-the-post system. Unevenness is a good thing.

    Ontario is the other Green battleground and the 12.6% here is very satisfactory. Last week's 9.3% is confirmed as anomalous, since all other recent polls were far higher. Keep your eye on Guelph, where Mike Nagy was 12% behind the winning Liberal in 2008. Successful entrepreneur, Guelph City Councillor and nominated candidate Bob Bell may be the first Green elected to the Hill.

    Don't Believe Any Single Poll. But there's no bad news for Greens in this one and a lot to smile at.

  31. Shadow: Eric the NANOS results are much more similiar to EKOS if you go into the data sheet and look at the second set of figures.


    Are we talking about the second columns under each region here? Because those are March results.

    Not that I don't prefer them, but that's not the hand that was dealt this month.

  32. Yikes John.

    Probably a bad idea to read something at 4 in the morning, should have looked at that with fresh eyes.

    The small print was saying something way different to my sleep deprived brain.

    My mistake, thank you for the correction !

    I always make an effort to accept whem i'm wrong, its part of a give and take !

    Thanks again, just glad to be valuable part of the commentary on here even if I do misread things now and then.

  33. I'm curious about the Bloc Québécois so-called "die-off" mentioned in a few comments in this thread.

    That's not what I can detect in either polls featured in this post (and BTW, you can't draw much conclusions either way with the Nanos polls, since their samples are so small), or in the multi-year trend lines and HoC standings.

    Peter, above, was right on the mark when he said "No Change". Millions of dollars have been spent and a lot of electrons have been spilled to endlessly report and comment on what amounts to a whole lot of statistical noise.

  34. I'm not quite sure why they are saying that either, the Bloc has been the most successful party in the House of Commons since 1993.

  35. Éric: I'm not quite sure why they are saying that [the Bloc will die off] either, the Bloc has been the most successful party in the House of Commons since 1993.

    Wishful thinking. Sort of like the Green Party getting 4% in the next election.

    “Close you eyes and tap your heels together three times. And think to yourself, there's no place like home.”

  36. Anyone else a little freaked out by the big stock market drops today, the big drop in the dollar ?

    Europe could be heading for a double dip pretty quick.

    Canada's exposure is limited but it would probably stall our recovery.

    Not sure who that helps or hurts. (Would normally hurt the party in power but Harper's an economist and polls show he is trusted on these matters.)

    However, its unlikely the governemnt would want to go into an election where they cannot control the narrative because of outside events.

  37. The Bloc will eventually die off. They lost alot of support last year when Michael Ignatieff first became leader, so if another leader comes along Quebeckers like it could be trouble for the Bloc and when Duceppe leaves they may also lose support.

  38. Actually it is only the beginning. The real "pigs" story isn't out yet.


    So far only Greece has essentially gone bankrupt. The rest are rocky. One more goes and there will ba cascade. Harper and every other pol will get burned as our recovery returns to recession. Make no mistake here, it will

  39. I suppose anything's possible, but it's about as possible as the Liberals dying off and being replaced by the NDP, like what happened in Britain with the Liberals and Labour.

    The best chance of the Bloc "dying off" is with Quebec separating. That is the more likely thing.

  40. The Bloc won't die off until one of the federalist parties is seen as fighting for Quebec's interests, which is almost a contradiction in terms.

    The closest the Bloc came to oblivion was when Chretien was leader in the latter years. Martin was even more popular, due to his past positions. But until there is a leader in that mold, either Liberal or whatever, then the Bloc aren't going anywhere.

  41. Oblivion is a little dramatic, considering the Bloc still won a majority of the seats in Quebec (38 in 2000) and has never gotten less than 37.9% of the vote.

  42. Volkov are you watching the British returns ?

    Lawrence Martin was writing today that the NDP and Jack Layton are all over the LibDems.

    Specifically this notion of a major party collapse and a third party replacing it.

    I've been talking about a convergence for awhile now, where the NDP-Liberal spread is within the MOE of a major pollster.

    The result would be similiar speculation and create a couple media cycles of Layton hype.

    If Elizabeth May was the media star last election, its possible that it could be Layton next.

    However! BBC's exit poll actually has the LibDems DROPPING 3 seats!

  43. BBC exit polls

    cons 305 +95
    labor 255 -94
    libdem 61 -1
    other 29
    conservatives 21 short of majority

    first three seats = labour, but a 5% swing in voters from labour to conservative. Libdem have not capitalized on any swing so far.

  44. In the Nanos poll, I don't understand how the Greens mathematically can have 0.3% of Quebec's vote. There were 198 people surveyed. If zero people said Green, then the percentage would be 0.0%. If one person said Green, the percentage would be 0.5%. How do they come up with 0.3%??? A bit fishy, if you ask me...


    BBC, if anyone else wants to watch

  46. UK Election

  47. A breakthrough in the UK election.

    The UK Green share of the popular vote is much smaller than the Canadian Green vote. The European Parliament uses proportional representation so Brits are used to having Green representatives (including Caroline Lucas), but this is still historic.

  48. There is a poll out showing the difference between RoC and Quebec....

    The nation in a nation malarkey again.

    In light of Greece with only 11 Million people, and following many of the same policies of Quebec, bringing the world to the brink of another economic meltdown will the world allow Canada to let Quebec go it alone.

    Socialism is great until you run out of other people paying for it.


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.

Note: Only a member of this blog may post a comment.