Friday, December 31, 2010

Two Years of Polling

As my regular readers know, ThreeHundredEight has now been in operation for over two years. The site began just shortly after the last federal election in October 2008. I think 2011 will be an exciting year in politics, and it will also be an exciting year for ThreeHundredEight.

Since February 2009, I've been averaging out polling results from each month. I've recorded them in a chart that only included the averages at the national level and for Ontario and Quebec. With today being the last year of 2010, it seemed fitting to expand that chart to include all six regions of the country. For simplicity's sake, I've also rounded all of the averages to the nearest full number.

The chart below tells the story of Canadian politics since January 2009. Click on it to see it magnified.This past year has featured incredible stability. The Conservatives have only fluctuated between 32% and 35%, while the Liberals have been between 27% and 31% and the NDP between 15% and 17%.

But 2009 had huge variations, the Conservatives polling between 32% and 39%, and the Liberals between 26% and 35%. The NDP were relatively stable.

Regionally, the greatest variations appear to have taken place in Ontario, Quebec, and Atlantic Canada.

In Ontario, the Liberals held the lead between January and August 2009, only to lose it and see the Conservatives roar ahead to more than 40% between September and November. Since then, the two parties have been trading the lead back and forth.

In Quebec, the Liberals were very competitive for most of 2009, faltering in October. They even came within three points of the Bloc Québécois in May 2009. The Bloc has been steady, while the Conservatives have also been steady except for the blip in support from October to December 2009.

In Atlantic Canada, the Liberals lost their lead in October 2009, regaining it in January. It was a close race until this past July, but the contest is getting closer again. You can also see that the NDP has fallen away steadily since last December.

I will update this chart regularly. It acts as a great look at how things have been going over the past 24 months.

All the best to you and yours for the next 12 months. I hope the year will bring happiness and success to all!


  1. The graph really hammers home how much more consistent and better the Green does in Polls rather than actual elections.

    Is this anomaly enough to invalidate all the polls? or just the Green/Undecided/refuse to answer portion?

  2. BC Voice of Reason: The graph really hammers home how much more consistent and better the Green does in Polls rather than actual elections.

    It's a bit difficult to follow the perceived percussion since there have been no general elections since Éric started publishing his graphs. There has also been very limited national support for Green GOTV (get out the vote) campaigns in by-elections and several of those by-elections have had star candidates (hello, Julian!) running for other parties.

    What happens when the Green GOTV resources are mobilized? Toronto Centre in 2008 is one example, and that against one of the biggest names on the Hill.

    The difference between the polls and past elections reflects GOTV effectiveness. Greens have steadily climbed in each election, but the increasing appeal of the message has been at least as important as the maturity of the political machine.

    That was then. I'm really looking forward to the next writ drop.

    Happy New Year all! (Non-political) party time now...

  3. Thanks Eric, that's a great graph, if for no other reason that it really shows the big picture. What's remarkable is it looks almost exactly like the same graph for 2007 and 2008 up until the 2008 election. Well, almost. Liberal support over the past year has fallen by about 2% compared with the 2007-08 period (which is an unflattering commentary on the Ignatieff-era). Anyone pondering an election in 2011 may want to ponder that little tidbit.

    Anyhow, thanks for all your hard work on this site over the past 2+ years. Happy new year.

  4. Seriously John... you are using a positive indicater of future success as a 13% Green GOTV in a riding that was a foregone conclusion. Then in the same riding with an experienced Green GOTV dropped down to 12% in the general election.

    The polls have the Green at a base of 10%. 12% in a good Green riding ranks as a disaster. The Liberals are at 26-29% and in a good riding (such as Toronto centre) they get 60 and 53%)

    Last election the Green got treated as a bona fide party based on the 10%+ they were getting in the polls and a floor crossing MP.

    They will have a tough time hanging on to that status next election.

    In Winnipeg North the Green dropped to .7% The Pirate party was within a handful of votes.

    The polls that offer Green as a valid choice are not as accurate as the polls who ask with party will you vote for (without a prompt).... just makes common sense.

  5. But is not supported by the facts. A pollster that ran both prompted and unprompted polls in the last election told me privately that the prompted poll was more accurate.

  6. "But is not supported by the facts."

    The experience of a single pollster making a comparison during a single election hardly settles the matter of prompted vs unprompted polls.

    In the last two elections Green support has been overestimated in polls.

    Especially so in the last election:,_2008

    I make no judgment either way. But if the underlying fundamentals are the same as in 2008 we can know this:

    CPC support is underestimated.
    Green support is overestimated.
    Liberals are in a weaker position.

    And the CPC is heading for a majority.

    But there's no way of knowing in any case until the actual election. And elections CAN change things.

    Also past events don't predict future events. SO as John said, happy new year all.

  7. It certainly does not settle it, but it gives us something to go on that's a little better than "common sense".

    In any case, the projection takes into account these over/under estimation factors.

  8. John said: "It's a bit difficult to follow the perceived percussion since there have been no general elections since Éric started publishing his graphs."

    True, but what's remarkable about the Green numbers since the 2008 election is how consistent they are with their polling numbers over the course of 2007 and 2008. And we do know how those numbers translated into actual election support for the Greens (I.e., they badly overstated it).

    John said: "The difference between the polls and past elections reflects GOTV effectiveness."

    Maybe, although you can't rule out the prospect that people (in particular non-voters) are simply parking their vote with the Greens. It's nor unrealistic to expect disenchanted Liberal/NDP/Tory voter to voting green as a form of protest vote against their own party (yes, even the Tories - I voted for the Greens in the 2003 Ontario election because I couldn't bring myself to vote for a real party). For example, was the "high" Green support in 2008 because of the popularity of the Greens, or was it inflated by Liberals who were underwelmed with their party but couldn't bring themselves to vote for the NDP or the Tories. That sort of "support" doesn't really reflect support for the Greens (in fact people vote for them because they know they'll never see the inside of the HoC, so they won't do any damage) so much of a loss of confidence in the other parties

    In any event, I don't see any reason to expect the green GOTV efforts in the next election will be more effective than they've been in the past. If anything, I'd expect them to perform worse than in 2008 given (a) the environment has fallen off the table as a significant issue, (b) the Liberals aren't so stupid as to endorse them twice (I hope), and (c) Liz May isn't going to get the coverage she got last time (either in quantity or in its uncritical character).

  9. Hey Eric

    It is interesting the spin that has become prevalent that the CPC do not want an election due to the polls.

    The last election was called Sept 7,2008

    The 13 polls from July 10 - sept 4, 2008 averaged CPC 35 Liberal 30 NDP 16 Green 9.5 and Bloc 8.3

    The CPC are at the same level they were when they called the last election and the Liberals are definitely down from their levels.

    There are definitely 20 Liberal seats that the CPC have targeted.

  10. Two question.

    Given that Crop has a better read on PQ(personal opionion),why do not not include the Crop poll results for Quebec in your averaging.

    The second involves the accuracy of the polling companies to their projections to actual results and do you make any allowances for the discrepancies that occur in your calculations.

  11. Eric

    What do you think the Pirate party would poll if it was added to the choices of parties?

    My opinion is that if the list of voting options was Pirate and Green the Green would be dropped to 3-5% and statistical tied with the Pirate Party. If Pirate was added and Green removed the Pirate party would be 8-10%.

    If pollsters believe that giving a list gives more accurate results then they should have no trouble adding Pirate and Libertarian to the list for increased accuracy.

    They won't. In this case they have
    common sense.

  12. Brusmit,

    CROP is included in my averages. I haven't added their latest yet.

    Yes, accuracy is taken into account. It is one of the three factors (sample size, date, and accuracy) taken into account to weigh each individual poll.

  13. But is not supported by the facts. A pollster that ran both prompted and unprompted polls in the last election told me privately that the prompted poll was more accurate.

    So you are saying that the Green polling went up without the prompting?

    That would have had to be the case as your pollster (as had pollsters except AR) had over estimated the Green vote and increased accuracy would have them down .. closer to the 6.8% they actually got.

    I call shenanigans!!

  14. "In any case, the projection takes into account these over/under estimation factors."

    Somewhat but you include the last three elections when calculating the strength of these factors.

    Only 2008 had the very strong variance.

    So as I said if the same dynamics are in play as in 2008 we can expect the Greens to be lower than in your projection and the Conservatives to be higher.

  15. BCVoR,

    I'm saying that the prompted polls were closer to the actual result than the unprompted polls. I don't understand what you're saying.


    2008 was only one data point. It is unlikely the exact same thing would happen again.

  16. "2008 was only one data point. It is unlikely the exact same thing would happen again."

    Oh sure the trend (of gradually increasing variance over the past three cycles between predicted Green support vs election day Green support) could continue.

    In which case the polls and your predictions would be even more off.

    But I don't think so. Everyone has commented on just how static the polls seem. Just how frozen in place the political dynamics in Ottawa are.

    I wouldn't be surprised if a 2011 election had the same polling variance as the 2008 one did.

    Everything else is essentially the same anyways except for the Liberals being slightly weaker and the Conservatives being slightly stronger.

  17. "And the CPC is heading for a majority. "

    And you really need to stop smoking that stuff !!

    As said elsewhere your continual pushing of the CPC makes a mockery of what Eric does !!

  18. Shadow,

    In fairness, the Liberals did considerably worse in the 2006 election than pre-election polling suggested and modestly worse in 2004. Over the past 50 odd years, the Liberals have generally performed about 4% worse in the election than the last pre-writ polls would suggest, which is approximately what we've seen over the past 3 elections, so Eric's correction shouldn't be too far off. And given the very large difference between Tory pre-writ polling numbers and the actual result in 2006, it's entirely possible that his method over-states the "over-performance" of the Tories relative to their pre-writ polls (historically it's been about 1-2%).


    What Eric is saying makes sense. After all, the only pollster to nail Green support (and the overall election results) in 2008 was Angus-Reid, which apparently prompts. Nanos doesn't prompt, and it was a close second. Of course, Ekos didn't prompt and Strategic counsel did, and both their predictions of Green support was well off the actual result, so it isn't entirely obvious how much significance we should assign to the prompt/don't prompt distinction rather than other methodological differences.

    Of course, the 2008 election is only one datapoint, so it's hard to draw too many conclusions from it (though, for the record, the closest pollsters in 2006 was Ekos, which didn't prompt, and Ipsos, which does).

  19. Carl you're talking about historical averages, which is what Eric was talking about.

    That's fine, as I said we have no way of knowing the magnitude of the polling/performance variance.

    Its possible the variance could increase, decrease, come in at some kind of historical average, or be a repeat of 2008.

    If, like myself, you believe that the underlying dynamics are the same as 2008 then as I said polls/Eric's projection will likely understate CPC performance and overstate Green performance.

    Conclusion - people who rule out the possibility of a CPC majority are very foolish indeed.

  20. What else can one base predictions on but historical performance? That's all we the data we have. Anything else is just a wild guess/wishful thinking.

  21. It would also be an interesting exercise to see this compared to a graph of past seat projections as well.

  22. Carl assuming that the variance is non-random and results from the nature of the current batch of pollsters themselves, the position of the parties, and or some kind of public sentiment then using a historical average might not be ideal.

    Pre 2006 polling had much lower green numbers, featured a Martin government, and a Liberal party that had not yet collapsed.

    It is similiar to pre 2004 polling and featured a similiar level of variance over both cycles.

    However, in 2008 there was a clear change, both in the nature of the polling/election variance, its magnitude, and of the political dynamics at play.

    We constantly hear how nothing has changed don't we ?

    Then why should we assume the variance has ?

    (Which is what a historical average assumes since it results in a much weaker Green reduction than 2008.)

    Using 2008 figures may be more exact than using '08/'06/'04 figures as Eric does.

    As I said, no way to know before an election.

    Pretending that one method is better than any other wat this point would be a wild guess/wishful thinking.

  23. Interesting poll results:

  24. Green support is a lot more 'soft' than other parties, in part due to the party being younger (in the respect of being taken seriously) and due to the ability for other parties to co-opt key Green planks.

    Last election one of the top Green planks, a tax shift from income to carbon, was partially stolen by Dion & the Liberals. Many Greens felt it was better to vote for the Liberals and try to get that policy put into place, thus would've said "I'm voting Green" but checked off the Liberals in the booth.

    The next election will not have that happen. The Liberals are pro-oil sands (among other non-green policies), as are the Conservatives. The NDP is all over the map (as per their usual). If you are a pro-Green voter you won't have an obvious place to 'strategically vote' in 2011 (or whenever the election happens) unless a drastic change occurs.

    Still, the GOTV efforts are becoming more and more important as people are turned off politics due to the ugly nature becoming stronger and more obvious. The big 3 have extremely strong GOTV teams (helps to have millions of government cash locked in thanks to the 60% refund they get for expenses among other things) while the Greens can't afford as strong of one. It will be an interesting election, that is for sure.

  25. Eric

    308 is very well done, far more effective than any other and much appreciated. I know the amount of effort it takes to produce the results that you produce.

    I sincerely hope you have found a way to make it pay. I say this because I hope to have your valuable resource available for decades to come.

    Great job ... please keep it coming.

  26. The consistent pattern in Eric's monthly polls and seat projections in the last year is that the libs will gain seats and the cons and ndp will lose some. The interesting question becomes how many will change hands and will the cons keep enough to form another minority. Is it worth it to have an election? I'm thinking not, although the cons are not good for the country. It might be better to wait until next fall when the government will die a natural death and the parties can campaign on real issues.


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.