Wednesday, March 2, 2011

Conservatives poised for majority in new Ipsos-Reid poll

The latest poll from Ipsos-Reid for Postmedia and Global TV is a humdinger. Normally, I wouldn't use such saucy language to discuss a poll, but it would appear to be appropriate in this case. The Conservatives have not only surpassed the magical 40% mark but have flown right by it, standing at 43% and a certain majority.

However, much of this Conservative gain could potentially be the result of a change in questioning method. And that's your anti-climax for the day.
Yes, compared to Ipsos-Reid's last poll conducted between February 8 and 10, the Conservatives have gained four points. The Liberals have also improved, picking up two points to reach 27%. The New Democrats have taken the brunt of the Tory onslaught, and have dropped five points to only 13%.

The Bloc Québécois is up one point to 10%, while the Greens are down five points to 5% in this telephone survey.

Woah, what happened to the Greens? Good question! According to their press release, Ipsos-Reid has now opted not to prompt for the Green Party. Respondents who say they will vote for the Greens are recorded, but no longer is the Green Party listed along with the Conservatives, Liberals, NDP, and Bloc. This is a major change, and is likely responsible for most of this Green slip.

But then where do their five points go? Clearly, they have not gone to the New Democrats. They also do not seem to have gone to the Liberals, who have only picked up two points. Potentially, three of those Green points have gone to the Conservatives, either directly or indirectly.

So what? Well, we don't really know whether prompting for the Green Party is more accurate than not prompting for the Green Party. One pollster I've talked to has told me that in 2008 their firm conducted their polls using both methods, and found the prompted polls to be more accurate. But it could have been a coincidence or luck. Either way, it casts a bit of a shadow on this poll for the Conservative result.

Otherwise, it is a stellar poll for them, though it would appear that only the NDP and Green drops are statistically significant.

One of the main sources of the Conservative gain seems to be among women. Though the gap between them and the Liberals in this demographic has only grown by two points, the Conservatives have increased their support among women from 34% to 41%, giving the party the rare feat (for them) of having almost equal support among both sexes.

Regionally, the Conservatives are leading in Ontario with 45% (+3), followed by the Liberals at 33% (+1). The NDP is down one point to 14%, while the Greens drop seven to 4%.

In Quebec, the Bloc is up four points to 41%, followed by the Liberals at 27% (a statisically significant gain of 10 points). The Conservatives are down two to 19%, while the New Democrats cruelly fall from 15% to only 6% in the province.

The Conservatives are up nine points in British Columbia and lead with 48%, while the NDP is at 22% (-8) and the Liberals are at 21% (-2). The Greens have dropped three to 8%.

In Atlantic Canada, the Conservatives lead with 52% (+14), followed by the Liberals at 33% (+2) and the NDP at 12% (-6).

The Conservatives lead in Alberta and the Prairies with 68% and 59%, respectively. That's a 16-point gain for the Tories in the Prairies. The Liberals stand at 17% in both regions, while the New Democrats are at 17% in the Prairies.

ThreeHundredEight's projection model gives the Conservatives a majority government with the results of this poll. The Conservatives win 25 seats in British Columbia, 28 in Alberta, 23 in the Prairies, 59 in Ontario, eight in Quebec, and 20 in Atlantic Canada for a total of 164. That's a gain of 15 seats over the last projection for Ipsos-Reid.

The Liberals win three seats in British Columbia, none in Alberta, two in the Prairies, 32 in Ontario, 18 in Quebec, and 10 in Atlantic Canada for a total of 66, a drop of 11 seats.

The Bloc Québécois wins 49 seats in Quebec.

The New Democrats win eight seats in British Columbia, none in Alberta, three in the Prairies, 15 in Ontario, none in Quebec, and two in Atlantic Canada for a total of 29 seats, a loss of one.

Note that the new projection model is completed, and that these projections were done at the riding level. As you can see at the top of the page, I will be updating the projection later today, with full riding details.

This is the sort of poll the Conservatives want to see - or do they? We've often seen the Tories drop after breaching the 155-seat mark. This is clearly a good poll for them. They are strong in every part of the country except Quebec, and even there they put in a respectable performance.

But this is a big number for the Conservatives. We will have to wait and see if other polls put the Tories in majority territory before we can start thinking that they are actually there. With the margin of error this Ipsos-Reid poll is well within the recent numbers we've seen from Abacus and Angus-Reid, but they are at the upper limit.


  1. I'm happy to see a minor nuisance party like the Green party finally being "de-listed". This is long overdue! Its worth noting that it was Ipsos that first essentially "created" the Green Party back in 2004 when they were the first polling company to start listing them - after being furiously lobbied by former Green leader Jim Harris. Before that the Greens were just an asterisk in polls, but of course once you start reading a party's name in a survey - you are juxtaposing them with the major parties. Imagine if the the Christian Heritage Party managed to get a major polling company to start reading their name in federal vote polls - they would probably instantly jump from .01% to 5 or 6% just as a result of being prompted.

    Hopefully Ipsos is the first in a series of dominoes as other polling companies de-list the Greens and we can concentrate our minds on the parties that matter.

  2. A shame that they decided a party that gets almost a million votes isn't worthy of being part of the poll anymore. Guess 900k+ Canadians don't count.

    Btw, last time I checked virtually every ballot in Canada for the past few elections has had a Green Party candidate, and next election will be the same. So on election day people WILL be prompted on voting Green - the only party that has A) released its platform (costed out as well) and B) actually wants to CUT payroll taxes rather than skyrocket them.

  3. All the pollsters need to take a much closer look at their methodology and consider what they are doing and how to be more accurate.

    If the pollsters were all doing a good job, they would be much. much closer together in results based on their statistical measure of error. But this is not the primary margins of error that should be considered, pollsters should be consider reporting their margins of error based on how close they are in their sampling to the typical norm of the Canadian voters. They should then also report on their known level of inaccuracy in polling.

    Part of the problem is that pollsters weoght based on empirical data based on the census. They do not weight their results trying to represent the people that will be voting in the next election. In part we have no idea what the maker up is of voters in the last election with any provable accuracy and we also have no idea who will be voting in the next election.

    The result of this is that pollsters have a much larger margin of error in all the polls than their statistical measure of error indicates and polls from different companies during the same period can be dramatically different.

    The easiest way for pollsters to test their methodology is to try different ways of reaching people and different ways of questioning them and testing them against the % that say they will not vote in the election. If this number is not within in the mid 50s, there is something wrong with how they are surveying Canadians. In the case of Alberta, this number should be less than 50%. If this, the easiest of questions to get right can not be accurately done, there has to be some serious thought given to what the poll outcomes really mean

  4. PS: I would like to know two more things about the impact of the "de-listing" of the Greens by Ipsos. First of all, do they now prompt for "other" or not? In other words, do they ask "Would you vote Conservative, Liberal, NDP or for some other party?" (if other, which one?) OR do they ask "Would you vote Conservative, Liberal or NDP?" PERIOD with the respondent having to interject "I want to vote GREEN - you didn't read that option!!"

    Secondly, I would like to know what happened to the percentage of undecideds after the Greens were "de-listed". My hypotheses is that the DK/NA would probably go up, because when you prompt Green it becomes a cover for a lot of people who are actually just undecided.

    All food for thought...some of the regional numbers in this Ipsos poll are pretty bizarre like the NDP at 6% in Quebec when every single other poll has them in the mid teens. I wonder whether they accidentally "de-listed" the NDP in Quebec as well??

    I wonder if Ipsos will now also start to de-list provincial Green parties that are essentially moribund (ie: all of them)

  5. I guess that IR is concerned about being accurate in making predictions rather than driving an agenda.

    There is statistical proof that the Green party never ever reached the voting numbers any where near their polling numbers. Why pollsters were taking so long to adjust their polls to the facts is astounding.

    Now we were being led to believe that the environment and Global warming are no longer major voter issues but the Green was going double its support in the next election???

  6. What's up with the ridiculously large MOEs in this poll? MOEs don't usually get that high, do they?

  7. DL,

    The undecideds went from 8% to 10%, so not a huge change.


    The regional MOEs are well within the norm for a poll of 1,000 Canadians.

  8. this is one poll but these results are just one of many polling findings that have the conservatives on a definite upward swing - also we should take note that the latest,what could be construed to be, witch hunt topics -"scandles"- such as bev oda and latest election financing charges brought against the conservatives are not of any big concern amoung most of us voters - it appears that we are more engaged in actuality such as improving economy, the reduced deficit projection the fact that our loonie is no longer the puppet of the usa,we now see ourselves as winners and not side watchers....i personally think that a coservative majority government is well within the realm of probability

  9. Regarding the new methodology and the Greens ...

    Both promting for five parties and not prompting for any have their disadvantages. And both methods may tend to bias the results.

    But continuing to prompt for the biggest four parties but not the Greens seems to me to be a slightly worse choice than the other two methods.

    The Greens are still likely to get non-trivial media coverage in the next election and, as John Northey points out, a Green candidate will likely appear on most ballots across the country.

    I can understand what Ipsos is trying to do here, but I'm not sure they've made the best choice.

  10. Should the CPC actually go into election night with numbers like these and that is a big question mark, I would expect them to end up with between 15 and 20 seats in Quebec. Quebec will want to be represented at the cabinet table. Hence I would conclude that a CPC majority could easily stretch into the 174 seat range.

  11. "Should the CPC actually go into election night with numbers like these and that is a big question mark, I would expect them to end up with between 15 and 20 seats in Quebec."

    Actually they did get numbers in the low 40s in polls just days before the 2006 election and they ended up at 365 on election day. If people in Quebec thought it was so important have MPs on the government side, they had ample opportunity to elect a slew of Tories in 2008 - and they didn't.

  12. Recall that CPC behaviour prior to the 2008 vote seemed designed to drive away Quebec support.

    It probably wasn't intended to do that, but it sure looked like it.

  13. Incidentally, this is the first time the Conservatives have polled voer 40% since they prorogued parliament at the end of 2009.

    I'd say the year has been very successful for them. The bad news all at once, and the good news little by little. Machiavelli wins again.

  14. Eric: a few months ago you posted a chart showing how different polsters "favored" political parties compared to others. I can't remember how IR and AR fared although I beleive they favor the CPC. Do you have those figures handy?

  15. Pinkobme,

    Though they are still somewhere on the site in the archives the last time I updated them, I took the house effect numbers down from the front page as I don't feel they were very reliable. Too many variables at play.

    But Angus-Reid and Ipsos-Reid do tend to have higher Conservative numbers. But, of course, Angus-Reid was also the most accurate pollster in 2008.

  16. Thanks Eric, I don't trust my memory so that's why I asked. However,I don't suppose it would make much difference. This IR poll would still probably show a CPC majority -- or at least very close to one.
    I just knew someone would mention the AR result for the 2008 election. I'm no expert but I've always felt they lucked out in getting such accurate figures. They were definitely helped by the decision by many Liberal supporters at the last minute to stay home and not vote at all.

  17. I find it odd that when analyzing polls you wonder "where did [Green] votes go?" and determine that they didn't go to the NDP based on the topline numbers.

    They certainly could have gone NDP, but other NDP voters went away in greater numbers.

    Polls don't tell us WHO goes where, just that the sampled numbers indicate movement by someone.

  18. Frunger,

    Quite right. I didn't mean it in terms of Green voters opting for the Conservatives, but rather that the missing five points have to be re-distributed somewhere.

  19. I don't think it's implausible, though, that delisting the Greens could directly improve CPC numbers. Vacillation between the Greens and the NDP is a protest vote against the traditional party structure, sure - but vacillation between the Greens and the CPC is a protest, arguably, against petty politicking. If 'how election campaigns are waged' leaves a bad taste in your mouth, and you're not excessively political, Green (when prompted) is a decent way of saying 'I hate the partisan slugfests in Ottawa'. In the absense of the Greens, that vote could default to CPC on the basis of 'if it ain't broke, don't fix it'. Also if the coalition bogeyman still exists, a fair percentage of possible Green/CPC waverers might not realise that May threw her support behind the coalition too in 2008.

  20. If I recall correctly, AR was also the only pollster to poll on the Thanksgiving weekend immediately prior to the vote.

    Nik Nanos was quick to point this out.

  21. The "minor nuisance party" gets 5% without even being prompted.

    Why? Because they're a valid alternative to a Liberal Party that's lost in its own delusions of grandeur. I joined the Greens after being asked what the Liberals stood for, and couldn't answer the question. The Greens have an answer to that, and it's surprisingly similar to where the Liberals end up. We just know why we're there.

  22. Please return the full-length posts in the RSS feed. Thank you.


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.