Monday, April 28, 2014

Conservatives and Liberals tied in Ipsos poll

When Angus Reid Global reported last week that the Conservatives held a narrow two-point lead over the Liberals, many eyebrows were raised. And with good reason: the number of polls released in the first year of Justin Trudeau's Liberal leadership in which the Conservatives held a lead could be counted on the single hand of a clumsy butcher. But instead of the next poll to hit the wires refuting these findings, Ipsos Reid's poll for CTV/CP24/Newstalk1010 has corroborated the new state of affairs.

Now that a month has passed since its initial release, the price of "Tapping into the Pulse: Political public opinion polling in Canada, 2013" has been reduced to $9.99. With elections approaching in Ontario and New Brunswick, now is the time to review what happened in 2013 and how the new leaders of the provinces' respective Liberal parties fared. The ebook can be ordered in all formats from Gumroad, or direct from Amazon for your Kindle or Kobo for your Kobo reader.

Ipsos Reid was last in the field between February 14-18. Since then, the Liberals have dropped four points to fall to 33%, while the Conservatives have increased by four points to meet the Liberals in a tie. The New Democrats were unchanged at 24%.

The Bloc Québécois was up one point to 6%, while the Greens were unchanged at 3% and 1% of respondents said they would vote for another party. Of the entire sample, 16% were undecided (a drop of two points). 

None of these shifts, including that of the Liberals and Conservatives, are outside the margin of error of a probabilistic sample of this size. However, the Tory and Liberal swing is large enough - and backed up by the swing recorded by Angus Reid - to pay attention to it.

(As an aside: Tapping into the Pulse includes a complete record of polling data from most of Canada's major pollsters for 2013. Ipsos Reid is one of the firms included in that record.)

Among what Ipsos classifies as likely voters (those who say that nothing short of an emergency would prevent them from voting, a method I consider preferable to turnout models based on potentially out-of-date assumptions) the Conservatives inched upwards to 34%, while the Liberals and NDP were unchanged. Compared to the mid-February poll, that represents a three-point gain for the Tories among likely voters, and a drop of two points for the Liberals.

While that is a small blip, the Conservatives have increased by six points among likely voters since Ipsos's previous poll of January 31-February 4, while the Liberals have dropped five points since then.

The surge in support for the Conservatives recorded by Ipsos Reid was less dependent on an increase in one province alone, which was the case of Angus Reid. For example, the Tories were up by double digits in Alberta and Atlantic Canada, as well as more modestly in Ontario, the Prairies, and British Columbia.

In B.C., the Tories led with 41% against 27% for the NDP, 25% for the Liberals, and 7% for the Greens. While the shifts since February 14-18 were insignificant, the Conservatives have now registered increases in support in three consecutive Ipsos polls stretching backing to November 25-27. In all, the Tories were up 13 points since then, while the Liberals dropped 12 points over three consecutive polls.

The Conservatives were up to 63% in Alberta (a gain of 17 points over the last two polls), while the Liberals were down 23 points to 16%. This is likely more of a reset from an anomalous poll than anything serious. The NDP was at 16% support here.

In the Prairies, the Conservatives led with 45% (up eight points over the last two polls), followed by the Liberals at 31% and the NDP at 20%.

The Conservatives were narrowly ahead in Ontario with 36%, followed by the Liberals at 34% and the NDP at 27%.

The Liberals were leading in Quebec, however, with 37%. That is up six points over the last two polls. The NDP was unchanged at 28%, while the Bloc stood at 24% and the Conservatives at just 9% support.

In Atlantic Canada, the Liberals were at 54% (down eight points over the last two polls), followed by the Conservatives at 33% (up 21 points since mid-February) and the NDP at 10% (down 15 points).

With these levels of support, the Conservatives would likely win around 150 seats, with 111 going to the Liberals, 58 to the New Democrats, 18 to the Bloc Québécois (thanks to vote-splits), and one to the Greens. 

The strong showing for the Conservatives in the West puts them back into a position to flirt with a majority government. They can win many more seats than the Liberals thanks to their advantage west of Ontario.

But what would the Conservatives need to push that bar over the 169-seat mark needed to form a majority government? Let's start with this Ipsos Reid poll, and adjust the numbers by one point at a time in each of the regions. We'll take 0.6 points from the Liberals and 0.4 points from the NDP for every uptick.

By using this method, once we get to 38% for the Conservatives, with the Liberals at 30% and the NDP at 22%, the Tories reached the 169-seat mark (the Liberals take 94 and the NDP 50). So, they are still quite a long ways from a majority government.

The Liberals are even farther. Starting with the same assumptions but subtracting 0.6 points from the NDP and 0.4 points from the Conservatives for every point gained by the Liberals, the party surpasses the 169-seat mark at 42%, with the Tories at 29% (and 118 seats) and the NDP at 19% (and 26 seats). This starts with the Conservatives at a relatively high level and the NDP at a relatively low one, so it may be overly penalizing on the NDP, but nevertheless the relative efficiency of the Liberal and Conservative vote is clear.

These two polls by Angus Reid Global and Ipsos Reid might mark a significant turn in the Canadian political landscape. Have the Conservatives really recovered from their historic lows? Is Trudeau's appeal starting to waver? Is this a momentary blip or something real? While we now have a much better idea of what is going on than we did when Angus Reid first reported, we are still in wait-and-see mode to see whether this shift is going to stick. 

34 comments:

  1. A little, likely a bit, I think an article in the globe had it as Trudeau and the media spending too much time talking about dropping the F bomb and not enough time focused on actual important issues. I suspect a bit of a blip but that's just an assumption, the Tories had a lot of bad news recently too, but who knows how much that actually changes polling numbers....to answer your questions

    ReplyDelete
    Replies
    1. Was the F-bomb really an issue outside of media personal and the Conservative base? I think one of the Conservative leaning pollsters was stating that the F-bomb shows makes Trudeau not look like PM material and that is why is numbers are softening.

      Delete
    2. The F bomb wasn't an issue in so far as people will base their vote on Trudeau's use of expletives. What the F bomb demonstrates is juvenile behaviour. Most mature adults are able to control their inner monologue, Trudeau's inability to do so does not make him look statemanlike. The F bomb was a minor incident but, can you imagine if the Pope was in town and he used that language or the Queen? He would cause an international incident and Canada would become a global laughing stock.

      The F bomb could be over looked if Trudeau had policies or inspirational oratory but, as of today he has been woefully inadequate on these fronts. If the Liberals had a plan for Canada's future no doubt it would blow over but, without policy proposals the media is limited by lack of information on what they can report about the Liberal party, instead the F bomb becomes the narrative on the Liberal party.

      Delete
    3. Actually, the Pope did drop the F-bomb:

      http://www.telegraph.co.uk/news/religion/the-pope/10673813/Pope-Francis-drops-F-bomb.html

      Delete
    4. And, as an aside, I doubt very much that many people noticed that story. We watch politics closely so everything is HUGE but most people do not, and that is hardly something to get through the barrier of disinterest.

      Delete
    5. Eric,

      My writing was not clear. I meant can you imagine if Trudeau launched the F bomb in the presence of the Pope or the Queen. In any case thanks for the link I'm sure it will be an interesting read.

      Delete
    6. To be fair to the Pope, I believe he mispronounced a word in Italian that sounded like the F-bomb. I don't believe the Pope even knows how to speak english.

      Delete
    7. Read the article. Instead of 'caso' (case), he said 'cazzo' (an Italian equivalent of f**k, literally c**k).

      Delete
  2. No way the Bloc is going to win 18 seats

    ReplyDelete
    Replies
    1. Many seats were NDP-BQ races. If the NDP drops 15 points and the BQ increases by one, who would win those seats?

      Now, if your comment questions the likelihood of the NDP dropping by 15 points and the BQ increasing by one, that is an entirely different issue.

      Delete
    2. New Crop Poll out of Quebec has NDP leading and Mulcair best pick for PM.

      http://www.lapresse.ca/le-soleil/actualites/politique/201404/28/01-4761763-sondage-crop-le-soleil-la-presse-duceppe-devance-peladeau.php

      Delete
  3. Can Tories really win 34% of the vote (and the Liberals just behind at 33%) and come 19 seats close to a majority? I guess the Tory vote in Western Canada + rural Ontario is that efficient.

    I'm surprised the Liberals are tied with the NDP in Alberta, while way ahead of them in the prairies.

    ReplyDelete
    Replies
    1. Alberta is a unique political scene because of the dominance of the Tories. Generally speaking Liberals do better in Southern Alberta including smaller cities like Lethbridge and the NDP's base is in Edmonton and to a lesser degree Northern Alberta. What that means is the 30-40% of Albertans who are inclined to vote for a party other than the Conservatives must vote strategically in order to have a candidate other than a Conservative or PC provincially elected in their constituency. Depending on how things go over the next 18 months I think either the Liberal or NDP vote could halve depending on what party looks best able to be the alternative Conservative vote.

      Delete
    2. Big Jay,

      I looked at the details of the poll. Among voters aged 34 and above the Tories are at 36%. Since, these are the people who vote I think it goes some distance to explaining how the Tories can be 19 seats short of a majority with only 34% of the vote.

      Delete
  4. I was wondering if you could explain how they calculate their confidence on their margin of error, they normally say +/- 3-5% or something close to that effect....what does that come from and how do they calculate it?

    ReplyDelete
    Replies
    1. They have a paper where they describe their 'credibility interval'. You can find it here:

      http://www.ipsos-na.com/dl/pdf/knowledge-ideas/public-affairs/IpsosPA_POV_BayesianCredibilityIntervals.pdf

      Delete
  5. So in a completely non-partisan way:

    Is the IR filter of likely voters enough of an adjustment to counter their 2 pt under estimation the CPC votes they had in their last poll before the 2011 election or is the most likely CPC vote already 1 or 2 pts higher and according to your model closing in on the 38% they need for majority?

    I can really appreciate the effort IR is apparently making in adjusting their polling method to attempt to capture the actual result if a vote were to be held today.

    Futher: Any comment on the polls that had the Liberals winning more seats in BC than they ever had in the history of Canada.

    Even using the IR poll your model you have the Liberals going from their current 4 seats to 12 (unbelievable?)..... That is a huge gain for the Liberals.... Do they have 8 new valid candidates in the West that can get elected? If they don't have the people in place campaigning full time by now it boggles the mind on them getting elected against a CPC incumbent who is campaigning almost full time and delivering Federal pork.



    ReplyDelete
    Replies
    1. Just because a poll under-estimated a party by X points in one election does not mean they will do the same in a subsequent election. If it were that easy, predicting electoral results would be a cakewalk.

      As to BC, the Liberals could well end up back on top and winning lots of new seats. Or they could revert to their current holdings. Local candidates and organization are important, but they only go so far. The 58 non-Mulcair NDP MPs in Quebec are a testament to that, particularly those who displaced Conservative MPs "delivering Federal pork." If the Liberals end up with over 30% support in BC, they'll win plenty of seats.

      Delete
    2. I know Nate Silver did a break down last US election showing which polling firms missed the actual result and by how much...Forgive my laziness here but did you do the same for the last Canadian election?

      Delete
  6. According to my simulator and the 308 map, I've got:

    126 CPC
    105 LPC
    68 NDP
    8 BQ
    1 GPC

    By region, it gives:

    Atlantic:
    24 LPC
    6 CPC
    2 NDP

    Québec:
    35 LPC
    26 NDP
    8 BQ
    6 CPC

    Ontario:
    48 CPC
    34 LPC
    24 NDP

    Prairies:
    18 CPC
    5 LPC
    5 NDP

    Alberta:
    27 CPC
    1 NDP

    British Colombia:
    21 CPC
    9 NDP
    5 LPC
    1 GPC

    Territories:
    2 LPC
    1 NDP

    ReplyDelete
    Replies
    1. Free of charge Sanity check... your simulator, the 308 map and/or the polls are WRONG:

      Territories 2011:

      Liberals came in a close 2nd in Yukon.... as the incumbent.

      They came in a distant 3rd with 18% of the vote in the western Artic to the NDP

      In Nunavit the Liberals were a distance 2nd with 28% of the vote compared to CPC cabinet Minister Leona Aglukkaq who got 50% of the vote.

      what basis would you have to possibly predict that Leona Aglukkaq loses her seat?

      Same question to Eric..... how would the 1 or 2 people sampled from the North change the status quo in these ridings?






      Delete
    2. For projections in the north, I swing the 2011 results according to the national swing in the vote since then.

      Delete
    3. To answer your question BCVoR, if I make a quick and simplified version of my simulator, in 2011, the CPC gained around 4,75% and the LPC lost around 6,75%. They are respectively put by the polls at -15% and +12,25%. That is a differential of 37,25%, so any riding the LPC lost to the CPC by this margin or less is fair play.

      In Yukon, in 2011, the CPC gained only 1% while the LPC lost 11,25%. The CPC, compared to their national results, has a trend of 1 - -4,75%, meaning -3,75%. The LPC, on the other hand, has a trend of -11,25% - -6,75%, meaning -4,5%. The CPC is at -15% at the national level + -3,75% for a total of -18,75% while the LPC is at 12,25% - 4,5% for a total of 7,75%. If you add those coefficients to the 2011 results, you get 15% for the CPC and 40,75% for the LPC. Easy LPC win.

      In Nunavut, the one seeming to cause you problems, in 2011, the CPC gained 15% while the LPC lost only 0,5%. The CPC, compared to their national results, has a trend of 15% - 4,75%, meaning 10,25%. The LPC, on the other hand, has a trend of -0,5% - -6,75%, meaning 6,25%. The CPC is at -15% at the national level + 10,25% for a total of -4,75% while the LPC is at 12,25% + 6,25% for a total of 18,50%. If you add those coefficients to the 2011 results, you get 44,75% for the CPC and 47,25% for the LPC. Tight LPC win. In my simulator, the difference is actually 0,48%, an even tighter victory (as there are other coefficients I calculate which I haven't done here for simplicity's sake).

      So to dismiss out of hand that the LPC could win those ridings as ridiculous is, simply put, ridiculous in itself. And considering you qualify yourself as the Voice of Reason, maybe you should put some discernment in your writings.

      Delete
    4. In reply to your snark maybe you should consider that the the CPC dropping from 16.5 % to 9% in Quebec will have Zero impact in Nunavit and adjust your model for this reality.

      Delete
    5. BCVOR,

      Please understand that things change. Unexpected things happen. Results from one election can be completely overturned in the next. Seemingly secure incumbents can be defeated. A few examples from 2011:

      Lawrence Cannon, important minister in the Conservative government, took 33% of the vote in the 2008 election in Pontiac, beating out the Liberal candidate by 9 points. Defeated by 16 points with 30% of the vote in 2011 by the party that had finished with just 15% of the vote in the riding in the 2008 election (the NDP).

      Josée Verner, important minister in the Conservative government, took 47% of the vote in the 2008 election in Louis-Saint-Laurent, beating out the BQ candidate by 21 points. Defeated by 2 points with 38% of the vote in 2011 by the party that had finished with just 10% of the vote in the riding in the 2008 election (the NDP).

      Jean-Pierre Blackburn, important minister in the Conservative government, took 53% of the vote in the 2008 election in Jonquière-Alma, beating out the BQ candidate by 15 points. Defeated by 8 points with 35% of the vote in 2011 by the party that had finished with just 5% of the vote in the riding in the 2008 election (the NDP).

      So yes, Leona Aglukkaq and many other incumbent cabinet ministers could very well be defeated in the next election.

      Delete
    6. The CPC may be down from 16,5% to 10% in Québec and that may not have much of an impact on te outcome of the election, but it is also down from 38% to 33% in the Atlantic, from 55% to 45% in the Prairies, from 68% to 63% (though that won't have much of an impact either), from 46% to 41% in BC and, most importantly, down from 45% to 36%. The territories may not be polled individually, but considering the party is down everywhere, it is logical to conclude that it will be down there as well. I also pointed out that the LPC victory in Nunavut, in my simulator, was by less than half a percent, meaning that it is more like a coin toss than a fact, so I am not pushing Leona Aglukkaq aside definitely, I'm just pointing out that her seat is far from a done deal if the situation reamins the same.

      Delete
    7. "down from 45% to 36%" in Ontario... kind of important to point that out.

      Delete
  7. And in the end despite all the pontification here the election will shred all the thinking !

    ReplyDelete
  8. This poll was also taken the week that the media was saturated with reporting of Flaherty's funeral, so I guess this indicates that it's had an effect to the Conservatives advantage. It is a real change in opinion, but it could disappear as fast as it

    Too bad the poll focused on the non-issue of vouching during elections. It would have been interesting if it had asked whether people thought Flaherty was a good finance minister. The question is whether the warm feelings about Flaherty will extend to the Conservatives beyond a week or two.

    ReplyDelete
    Replies
    1. Guy I've been making that point for over a week but it is just rejected by the "convinced".

      Delete
  9. Peter: It's a legitimate possibility. But it looks for now that government coffers are overflowing as the price of oil increases and the Canadian dollar increases. That could help the Conservatives capitalize on this.

    It also shows that the support or Trudeau is rather shallow. I'm wondering if nostalgia built the Trudeau lead and is now eroding the Trudeau lead.

    ReplyDelete
    Replies
    1. The Canadian dollar is down about 2 cents since 2014 and roughly 8 cents over the past year. The trend is a downwards.

      Over the last six months oil is up very slightly about a dollar a barrel. Perhaps Western Canadian Select (Tar Sand Crude) has had a greater appreciation but, usually it sells at a discount to West Texas International the North American "oil standard".

      Regardless of it being a "Flaherty bump" these polls clearly show the shallowness of Trudeau's support especially in Ontario.

      Delete
  10. The dollar is down and so is oil. I agree though that this is a reaction to the death of Jim. A lot of small c cons were feeling nostalgic. In two weeks they will remember that harper and hudack don't directly answer any questions and will see Rob ford part 2.

    ReplyDelete

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.