Saturday, October 10, 2015

2015 Federal Election Link Round-up, Week 10

The debates are done and we now find ourselves in the penultimate week. A decisive week?

Saturday, October 10, 2015

- The Poll Tracker and riding projections have been updated.

- I talked to Chris Hall on The House about the TPP and the trend line heading into the final week.

- Another four polls this morning: EKOS, Nanos, Angus Reid Institute, and Innovative. Innovative and Nanos show the Liberals moving ahead, and EKOS has it tightening up. ARI has moved into agreement with the other pollsters compared to where it stood before.

- Some long-weekend scheduling: I'll be updating the Poll Tracker today and on Monday, but not Sunday.

Friday, October 9, 2015

- The Poll Tracker and riding projections have been updated.

- The latest episode of the Pollcast, with unofficial co-host David Coletto of Abacus Data. We talked the change vote and the TPP.

- Looking for a poll today? Looking for four? Nanos, Léger, EKOS, Mainstreet. All but Mainstreet show the Liberals ahead, and even that poll has the gap at one point after putting it at seven.

Thursday, October 8, 2015

- My latest analysis of the new polling numbers.

- The Poll Tracker and riding projections have been updated.

- Talking about the polls on last night's Power and Politics with Rosemary Barton.

- The latest episode of the Pollcast, featuring Susan Delacourt! We talk about what impact the polls have had on this campaign, and how the media has been covering them.

- Les tendances des sondages - Matins sans frontières, Radio-Canada Windsor.

- The last 18 hours of polls: Nanos showing the Liberals down, the NDP up, and the Conservatives up a little since their previous three-day sample. Since a day before, EKOS showing the Tories and NDP down, the Liberals up. And Forum showing the Liberals up and the Conservatives down from their previous survey. A lot of convergence, then, from the more confusing picture the polls were painting over the past week or two. Also, some riding polls from Mainstreet for the Ottawa region.

Wednesday, October 7, 2015

- The Poll Tracker and riding projections have been updated.

- My latest regional spotlight is on Toronto.

- Your daily Nanos. Compared to the previous three day sample, every party has moved about a point: the Liberals and NDP down, the Conservatives up. And an Abacus Data poll just released, showing the Conservatives up very slightly, and the Liberals making a more substantial gain at the expense of the NDP. With this poll, and the EKOS poll yesterday, I think we're starting to see more convergence.

Tuesday, October 6, 2015

- My latest analysis of the polls, and how they are still in disagreement. And this disagreement might continue on a daily basis, as EKOS will now be releasing daily tracking polls too!

- The Poll Tracker and riding projections have been updated.

- A few national polls released in the last day. Your daily Nanos, which compared to the previous independent three day sample is showing the Liberals and Conservatives up very marginally and the NDP down less marginally (and overall, the Liberals leading the Conservatives by 3.5 points). Ipsos Reid is showing very little change from their previous poll, with the NDP and Liberals down a point and the Conservatives up one (the Conservatives lead the Liberals by one). And then there's Mainstreet, which shows the Conservatives almost in majority territory (the Conservatives lead the Liberals by eight). Their previous poll was done almost two months before, so the trend line is not very informative. Also note that Mainstreet was out of the field on October 1, whereas Nanos and Ipsos were out of the field on October 5.

- Another bunch of riding polls from Forum. Check out this page for the full list.

Monday, October 5, 2015

- The Poll Tracker and riding projections have been updated.

- Your daily Nanos. Compared to the last independent sample from Nanos, we're looking at a small drop for the Conservatives (0.9 points), a decent increase for the Liberals (2.1 points), and a slide for the NDP (3.1 points). Compared to the sample before that, we're looking at much more substantial movement (-1.6 for the Conservatives, +4.2 for the Liberals, and -4.2 for the NDP). Also, a riding poll from Forum for Peterborough–Kawartha.

- If you're up this early, you can catch me on CBC Quebec City radio (in English) at about 8:10 ET and on CBC News Network at 8:40 ET.

254 comments:

  1. Ok, so I looked at the Nanos regional numbers: Am I the only one who finds them a bit... odd? I already didn't like that daily rolling numbers thing, now those numbers... I don't know, it makes me hesitant to find them credible. Maybe I'm projecting personal bias, so I'd like other people's opinion here.

    ReplyDelete
    Replies
    1. What do you find odd?

      Regardless, Nanos has a very good track record so I don't think one could reasonably question the credibility of the firm.

      Delete
    2. I think the biggest issue with Nanos is their age-based weighting. From Elections Canada, the average vote by age range (38th to 41st) was much more skewed from the population distribution.

      The raw response rate was fairly close to the historic voting average, but then Nanos increased the "youth" vote to 20.0% of the total at the cost of the "senior" vote to 23.7%

      Historic average: 14.5%, (18-29), 15.5%, (30-39), 20%, (40-49), 20.2%, (50-59), 29.9%, (60+)
      Nanos unweighted: 15.5%, (18-29), 16.4%, (30-39), 16.8%, (40-49), 19.8%, (50-59), 31.4%, (60+)
      Nanos weighting: 20%, (18-29), 16.9%, (30-39), 21%, (40-49), 18.4%, (50-59), 23.7%, (60+)

      On a related note, could the "shy tory" effect, really be a by-product of skewed age weighting?

      Delete
    3. They are odd. But the margin of error is massive for regional numbers. How many people are they interviewing per night in BC for example? 50 is my guess. So to have a swing so large in BC over a day means what? A bad news day for a party.

      Delete
    4. It doesn't even take a bad news day; Brownian motion will do it. For the regionals even more than the national numbers, I mentally hold a ruler across the entire reporting period. You can discern real trends over a somewhat shorter period, but any two-day shift is just noise. Regardless of your leanings, don't be too happy or too upset.

      Nothing wrong with the polling data; it's the interpretation of that data that matters.

      Delete
    5. 400 people across the country each night means that other than Ontario and Quebec, the sample size is likely too small to give much meaning to the swings.

      Quebec is really interesting because I think it is showing how badly Mulcair handled the niqab issue, resulting in a decline there with a commensurate rise in the BQ.

      Delete
    6. To Mapleson: YES! Absolutely. The way we poll in Canada without likely voters and without reporting the undecideds is completely unreliable. It's mostly luck that they get even close, well they modify their methodologies nearer to the vote.

      Polling is good for trends, and if you separate out the polling firms on the Globe graphs, you see for the most part the same trends in almost ever company, over time, though perhaps not at exactly the same time. It's fun you should try it.

      Delete
    7. I don't treat the Nanos numbers as gospel by any means, either, but most of them are close to what other pollsters are showing for most of the regions.

      The biggest difference seems to be in Ontario where Nanos shows a larger Liberal lead than is probably the case.

      On the other hand, the Nanos numbers for Quebec are a lot more credible than Ekos. Ekos has a probably with its Quebec numbers right now because it recently had the Liberals in 4th at 16% down at Ignatieff 2011 levels. This is in total contrast to the big Leger Quebec poll that had the Liberals at 24% (which is where most other pollsters are showing them).

      So each pollster seems to have its own issue with regional sampling. Nanos seems to be overestimating the Libs in Ontario by several points, and Ekos is underestimating the Libs in Quebec by quite a few points.

      Delete
    8. The nightly regional numbers are easy to tell, in the case of BC, it's 60 (180 total). The MOE is 7.7%, so the NDP is in a statistical tie for first and/or third.

      Delete
    9. Nanos's methods tend to be slow to notice changes to long-term patterns. If people have always voted Liberal, their knee-jerk reaction when asked is typically to say Liberal.

      Pollsters who offer prompts don't have this problem, because the respondent gets to head the list first, reminding him what his current preferences are.

      Delete
    10. "Likely Voters" have been less accurate than raw polling in the past. Nanos does report the undecided numbers. It's been hovering around 11%.

      @Paul, the proportion of the 400 for each region is in line with the population/seats. Statistically, the Atlantic is slightly over-polled then reduced down to get significant numbers out.

      @Craig, I've been looking at Nanos more just because they've been dominating the Poll Tracker due to their 3-day average releases.

      Delete
    11. Well Craig 2 days ago the Nanos regions were wacktastic but today they are moving back into reasonable. Liberals leading in Atlantic but Conservatives will do fine and maintain most of their seats. Quebec is still NDP but by less Liberals will gain seats, Cons still polling above 2011, we'll see if it last to E Day. Ontario competitive between Cons and Libs, (the difference in who is ahead here is the difference in the polls overall) Prairies are COn territory at least 50%, BC is and always will be until E day a 3 way race. Large swings in BC numbers should just be ignored, that's MOE sample is too small. Ontario difference is where the polls are defined.

      Nanos will continue to swing back into normal ranges for more days.

      Delete
  2. These polls are a tool that is well manged. I have confidence that in taking a number of polls we are not being manipulated by partisan olls. Thanks for this.

    ReplyDelete
    Replies
    1. Sort of. Nanos is given too much weight and It's arbitrary. It's just because Eric thinks Nanos should be given more weight. If you check the Globe aggregate, it uses standard statistical modeling and gets results that are not spiked so often by Nanos. It's a more stable model.

      Delete
    2. Nanos gets more weight because of the track record, not because of what I think of Nanos.

      Delete
  3. Eric, is CBC ever going to get the Nanos weightings right for 2-3 days in a row?

    Weighting of Nanos Polls going backwards:
    Oct 2: 25; 5; 0; 0; 7; 0; 0; 2; So polling for Sept 30 and Oct 1 are double counted, and Sept 29 is under counted.

    ReplyDelete
    Replies
    1. You're reading it wrong.

      The most recent poll gets full weight. The last two polls might get partial weights, depending on their place in the three-day pattern. Once we're far enough back to reach the last unweighted poll, we get two unweighted polls, and then one fully-weighted poll.

      It makes sense. The math works.

      Alternatively, Éric could count always the full weight of the most recent poll, and then the full weight of the last independent 3-day sample, but that would be a lot ore work because he'd have to redo all the past results with every new day.

      Delete
    2. I'm definitely not reading it wrong, those were the weightings before today's update. Today's update has every third Nanos poll with weighting and the intermediates with none. Why should intermediate polls be weighted at all? Especially the most recent two. Even then, if it were consistent for one day to the next, I'd not mind as much.

      Delete
    3. As an aside, weighting every third poll is just matter of setting the weight to an IF statement based on mod3.

      Delete
    4. Each independent three-day sample is given 100% weighting every three days, with intermediate polls being awarded 0%.

      But then for the next two days, I give the latest intermediate polls a 33% weighting to represent the one day of polling not represented in the newest set of numbers, but that is still represented in the intermediate polls.

      Delete
    5. This skews the Poll Tracker by unequal weighting between polling days. The current system meant that on the day after a three-day independent sample, older polling data has a higher weighting than newer data. Using the Oct 2 numbers before and assigning 1/3 of the weight to each polling day with the given weights you have: Oct 2 - 8.3%; Oct 1 - 10.0%; Sep 30 - 10.0%; Sep 29 - 1.7%; Sep 26-28 - 2.3%.

      Unless Nanos stops polling on a third day, this is going to show up in the final Poll Tracker predictions. It'll be interesting to see how much the final results are skewed by this (significant or rounding error).

      Delete
    6. Mapleson - Yes, but each new day would then change the historical data on which past projections were based.

      Delete
  4. The TPP could be interesting. Some parts of the country will probably hate it, and some parts of the country will probably love it.

    During the first English debate, Mulcair was going after Harper for failing to get any trade deals done. Now he has. I don't recall what Trudeau's stance on the deals was.

    ReplyDelete
    Replies
    1. Trudeau's general take on trade deals has been positive. He's used it in the debates as a point against Mulclair. For TPP specifically he's mostly criticed the lack of gov transparency on the details and choose to withhold comment until such details emerge.

      Delete
    2. Trudeau said the Liberals support free trade (but will try to make sure there are adequate safeguards in the deal).

      Delete
    3. "Some parts of the country will probably hate it,"

      The auto-area (AKA Ontario and a little bit of Quebec) will hate it, but fail to recognize that if Canada is not part of TPP then NAFTA may apply to cars coming into Canada, but we will basically be shut out of America since the content requirements will be below the current requirements.

      TPP is very bad in terms of copyright law and Internet surveillance, but I suspect few people will notice that issue.

      Delete
  5. The ABH force is strong. The tide is turning. The more negative Harper gets, the more desperate he will look. The only thing that could help him now is elections fraud.

    ReplyDelete
    Replies
    1. I don't expect negative ads.

      I expect ads that play on uncertainty. I expect ads that juxtaposition campaign statements from his opponents with contradictory facts.

      The Tory radio ads so far are quite positive.

      Delete
    2. "The only thing that could help him now is elections fraud."

      Err, no. While these may not be plausible, there are many things that could yet occur, not least of which would be an ISIS terrorist attack on Canada that was significantly more deadly than anything Canada has currently experienced. While I wouldn't wish it to occur, and have no love of Harper, I have absolutely no doubt that it would shift the polls dramatically in Harper's favour.

      Delete
    3. well insults seem to be the Liberal's purview of late. With Justin's opinion that the work of David Suzuki is nothing more than "sanctimonious crap". Very unbecoming of someone who wants to be prime minister.

      Delete
    4. That's just about the only thing Justin Trudeau has ever said with which I agree.

      Delete
    5. FCPete, Justin never called Suzuki's WORK sanctimonious crap, just his comments to him (Justin) on the phone. The comment was well-merited regardless of what you think of Suzuki's work, which I happen to value. Don't play so fast and loose with the truth, it's not helpful.

      Delete
    6. Thingamabob,

      Go you know what to yourself-How dare you insinuate I'm lying-shameful!. I'll let you know I have first hand information on this topic and while the comments may have directly referenced Dr. Suzuki's phone comments-they inferred judgement on Dr. Suzuki's work! The fact Suzuki felt the need to speak publicly about this says it all. Shameless of Trudeau-he's not even in Government and already in the pocket of Bay St. If you care about the environment vote Green a vote for the Liberals is a vote for pollution!

      Delete
    7. @FCP: Trudeau's exact words, as quoted in the National Post, were ‘I don’t have to listen to this sanctimonious crap.’ which is inline with what Thingamabob stated. As for the rest of your comment, you are simply placing yourself in the category of CPC partisan who desperately wants to split the vote and win based o FPTP voting system, which is also consistent with your other comment telling people not to vote strategically; sorry, but people can and will vote however damned well please.

      Delete
  6. BY the way Eric stating the Liberals are now leading in the poll tracker based solely on one polling firm Nanos is lacking in credibility and risks your reputation, given the last few other polls showed them behind by as much as 7 points. I thought you had a rule of two polls. There are not 2 full polls even from the same company that has made that shift. You should have waited to update the tracker. it looks like propaganda at this point.

    ReplyDelete
    Replies
    1. You are being rather selective in how you're reading the polls. Leger and Innovative also have the Liberals leading.

      Delete
    2. So... you think Eric should simply ignore all polls that show the Liberal Party in the lead and instead only include those that show the Conservative Party in the lead? I know confirmation bias is a hard thing to set aside but you're not even making an effort.

      What exactly do you think Eric's incentive is? Because as far as I can see he has zero incentive to be incorrect and plenty in being accurate. I get the impression that you believe he has some sort of clandestine agenda against your rooting interest... if true I think you need to give your head a shake and Eric should feel flattered that you believe he holds the fate of the nation at the tip of his keyboard.

      Delete
    3. Three of five polling firms show liberals out in front while the remaining two show conservative. I see no bias or credibility risk whatsoever.

      In saying that, I'm interested to see the next non forum poll as some of them are already a little bit stale dated.

      Delete
    4. As Eric says, Nanos is not the only pollster showing good numbers for the Liberals. Leger & Innovative had them ahead last week too, and today's new Ipsos shows the Libs only 1 point behind the Conservatives.

      Personally I think that it's a close race for 1st between the Libs & Cons right now. Liberals are probably not as high as Nanos says, but other polls are beginning to show them with decent numbers, too.

      Delete
    5. My point was about changing the Poll tracker for 1 and 2/3 polls from the same company. Eric has a rule about using 2 polls. This should not constitute 2 polls.

      Delete
  7. Eric, with the riding projection being somewhat tenuous, why do you express the percentages with one decimal place? Your explanation reads, "These riding projections are the best estimates of likely outcomes if an election were held on the last day of polling. But to give the Conservative percentage for Brandon, MB, as 49.1%, don't you imply that the number is between 49.0 and 49.2%? Also, in the Brandon case, have you ignore the 2013 by-election results? Why not report whole numbers?

    ReplyDelete
    Replies
    1. The polling data is usually released with one decimal place. The Min and Max give the range, while the precision is based upon the raw data. You can always do your own rounding.

      By-election results are included in the model.

      Delete
    2. Mapleson, the projections for Brandon-Souris show CPC with 50.6%, the LPC with 27.9%. 2013 by-election results had the CPC win the seat with 44% of the vote and the LPC candidate with 42.6%. However, the relatively poor results for the CPC candidate were in part due to some infighting within the CPC as that riding is traditionally a safe CPC seat. To what extent were the by-election results factored into the riding projections?

      Delete
  8. I'm curious how far the NDP would have to sink to put Halifax in play for the Liberals.

    ReplyDelete
    Replies
    1. Hard to say. NDP numbers appear to be dropping in the Atlantic since the bump they got at the start of the Summer.

      Megan Leslie probably still has an edge as an incumbent, but the Liberals gave former NDP Leader Alexa McDonough a competitive race in 2004 when they ran Sheila Fougere who placed a strong 2nd.

      Delete
  9. Eric, another question.The Liberal candidate for Victoria, BC, resigned recently (http://www.thestar.com/news/canada/2015/09/30/liberal-candidate-in-bc-resigns-over-facebook-posts.html). If there is now no Liberal candidate running, why do you still list the chances of a Liberal candidate as 24.3%?

    ReplyDelete
    Replies
    1. She will still be on the ballot, but this may be worth a look.

      Delete
    2. Eric, yes, you're correct in that she's still on the ballot. Interesting to see what happens to candidates who have withdrawn and are still listed on the ballot. There have to be other cases in the past.

      Delete
    3. I went ahead and did just that. Small sample size of recent examples to base it on, but I dropped the Liberals' support in Victoria to 39% of what it should be.

      Delete
    4. Thanks, Eric. Your website is fascinating and your response is commendable.

      Delete
  10. Based on the last three or four days of Nanos numbers I think the ABH vote is moving from the NDP to the Liberals. Quite possibly because Mulcair has taken an essentially left party to far to the right and possibly indeed past the real centre !!

    ReplyDelete
    Replies
    1. I think Mulcair overestimated the voters.

      He's presented a fairly sensible platform, but his party isn't historically known for that, and people are looking at the label rather than what's inside.

      Delete
    2. Half right Peter. The sad truth is nobody likes Mulcair. He could promise everyone puppies and free cash and Canadians would still dislike and distrust him.

      Delete
    3. @Ira: "He's presented a fairly sensible platform, but his party isn't historically known for that"

      I agree, though I think his failure to clearly and unambiguously slap down the Leap Manifesto crowd is a significant component of that problem. He needed to take that moment and use it to assert that this is not that party (in the same manner that Blair referred to "Labour" as "New Labour", though "New NDP" doesn't quite work so well ...).

      Delete
    4. He has a beard.

      Canadian voters don't like beards. When Garth Turner ran for the PC leadership, he should have won. But he was never a serious candidate, because he kept the beard.

      Delete
    5. @Ira: You may be correct. I have often thought that Joe Clark would have got a majority if his wife was called Maureen Clark and not Maureen McTeer (especially in a much more small-c conservative Canada as was the case in 1979).

      Delete
    6. I think it's just a matter of party branding. Party support looks a lot like it did 3 to 4 elections ago, with the exception of the Bloc -> NDP movement, and even there some of the people who left for the NDP last election seem to be coming back.

      The NDP are on track to their second-best result ever. It only looks bad by comparison to Jack Layton's win, and to their position at the start of the campaign, where everything from Notley's upset victory to the Duffy trial was in their favour. Nobody's going to get all of the bounces all of the time.

      In the US, it's conventional wisdom that people campaigning for the Presidency have advantage in their home states. Does anyone really think this doesn't apply in Canada, too? At the death of Jack Layton, NDP support in Quebec was a mile wide and an inch deep. Choosing Mulcair for their next leader gave them time to put down roots in Quebec. If they'd picked a leader from RoC, they might very well be losing all of their ridings in Quebec as fast as they gained them. But the other side of that coin is that it's not reasonable to expect the NDP under Mulcair to do dramatically better in anglophone Canada than the NDP under Layton did, and it's especially not reasonable to expect Mulcair to do better in RoC against the competent Trudeau than Layton did against Ignatieff, who was by far the weakest federal Liberal leader in my lifetime.

      As for the Liberals, they've averaged 113 seats over the last 6 elections. For them, this election is a reversion to the norm.

      Delete
    7. @ Paul A.S. Ward - A lesson Stephen Harper learned. As soon as he was CPC leader and striving to become Prime Minister, suddenly Laureen Teskey became Laureen Harper.

      But up until then, she'd always been Laureen Teskey.

      Delete
    8. Are you sure about that? My recollection was that Laureen Teskey kept her maiden name while her husband was Leader of the Opposition. It was only when she became the PM's spouse -- a role that defines a person much more than a party leader's spouse -- that she became Laureen Harper.

      See, for example, this G+M profile: http://www.theglobeandmail.com/news/politics/dont-call-her-first-lady-a-rare-sit-down-chat-with-laureen-harper/article9561792/?page=all "That was around the time she transformed from Laureen Teskey into Laureen Harper, a public move meant to defray any potential confusion now that her husband was Prime Minister."

      I can very much believe that Maureen McTeer's image as an independent woman hurt her husband politically, because it made him seem less of an alpha male, if his wife didn't even boast his name! But Canada has changed, the 2-name family isn't radical any more, and I can't see it as an issue.

      Delete
    9. Then I'm actually quite disappointed in her decision to change it. I actually oppose spousal name changes (I find it disturbingly patriarchal), so if she didn't need to change her name for reasons of political expediency then I wish she'd kept it in order to offer a prominent example to young Canadian women that their identity need not be tied to that of their spouses.

      Delete
  11. EKOS also now has released daily tracking with 3 day rolling averages that could be directly compared to Nanos, with appropriate weighting.

    ReplyDelete
    Replies
    1. Don't bother. The partisans are going to see what they want to see, regardless of the evidence.

      Delete
    2. The EKOS "3 day rolling averages" aren't released daily, and the weekly release only has one set of numbers. You'd have to guess everything else.

      Delete
    3. Frank Graves says EKOS data is now showing the Liberals moving into a tie with the Conservatives in Ontario whereas the previous EKOS poll showed the Liberals a number of points back in Ontario.

      While EKOS is still showing lower Liberal numbers than Nanos, it could mean there will be some convergence between the two pollsters soon.

      Delete
    4. Where do you find these EKOS daily tracking polls? I can't find them. Moreover, their last numbers out on October 2 were actually stale, since they cover September 23 - 29. They have repeatedly been outliers, but if this is because their weighting is more accurate, they could end up the most accurate on election day. Historically this has not always proven to be the case.

      Delete
    5. NO Mapleson if you scroll down on EKOS there are daily 3 day rolling averages. Yes they are released later but they could still be compared and given the same weighting.

      Delete
    6. Ekos is now a 3 day poll period and uses live interviews for 1/3 of the responses. I don't know if that is continuing or not, but it should now have the same weighting as Nanos or Eric is just being arbitrary and statistically questionable.

      Delete
    7. Polls are weighted by three factors: date, sample size, and pollster track record. There is nothing arbitrary about the latter weighting factor, as all pollsters are graded by the same measures.

      Delete
    8. Measures that you decide on are arbitrary Eric. People could and do disagree with your measures. I don't think online panels should be weighted at all for example. i don't think they are representative polls or have a margin of error and therefor should not be counted. You count them.

      Delete
    9. The Globe model that you seem to prefer also includes online polls.

      Delete
    10. shoshana, a 3-day rolling average is shown in graphic form, but the data is released on a weekly, not daily basis, and only the last 3-day period has numbers given.

      Just because you don't agree with Eric's system doesn't mean that his methodology is arbitrary.

      Delete
    11. And online polls have been more precise in a number of elections.

      Delete
  12. There appear to be recent Ekos and Forum poles that are not included here. Has Eric dropped them?

    ReplyDelete
    Replies
    1. No poll has been dropped, but weight decreases as time goes on.

      Delete
    2. Sorry Eric but Nanos carries way too much weight in your model, and is counted too often and really that's just arbitrary. It's not statistically sound. You just decided that's the way it should be.

      Delete
    3. Well compare your model to the statistically sound Globe and Mail model Eric. You are not a statistician. You are simply averaging polls with an arbitrary weighting system you have decided on you also decide when to update your model and poll projections at random times. It is questionable.

      Delete
    4. This comment has been removed by the author.

      Delete
    5. Shoshana - I update daily, so I don't know how that is random.

      Delete
    6. In what way is the Globe & Mail model more statistically sound than Éric's?

      If you think that Éric's model isn't statistically sound, you should be able to show that with math.

      Delete
    7. shoshana, if 308 is so questionable, why bother posting? The fact is that Eric's model and the G&M simulations are in line. They are different approaches to the same problem and both give additional clarity to the situation. The underlying assumptions for both are the same: individual polls can be biased and/or inaccurate; voting patterns shift on a regional level; and, polls are weighted by age decay.

      Really, there isn't anything stopping CBC from taking the model and running error tolerance simulations like the G&M. The question is just how much extra value that step really adds.

      Delete
  13. Eric, it appears my reply to "jimmythedeke" and "JF" in Week 9 comments has been over-looked. Why would this be so?



    ReplyDelete
  14. Tue Oct 6 Nanos

    LIB 35.0
    CPC 31.5
    NDP 23.1

    Little change from yesterday

    ReplyDelete
    Replies
    1. Wed Oct 7 Nanos

      LIB 34.3
      CPC 32.1
      NDP 23.0

      All twitches within the MOE !

      Delete
    2. Thur Oct 8 Nanos

      LIB 33.5
      CPC 31.6
      NDP 24.2

      Delete
  15. While Nanos shows a continuing close race between the Liberal and Conservatives I think it more important to look at his regional numbers to get a clearer indication of trending and poll standings.

    If, as most believe, this election's winner will largely be determined by Ontario and BC voters we see a distinct Liberal advantage. The October 6 Nanos numbers have them up by 7.5 in BC and 5.8 in Ontario.

    That said, the nature of this election leads one to believe even these numbers could be misleading as it really will come down to individual riding trench warfare and all that entails...including an unprecedented level of strategic voting by individuals of all political stripes and persuasions.

    ReplyDelete
    Replies
    1. The polling data suggests that Quebec is in play, and so it could be very significant in this election. If the Mulcair vote collapses in Quebec, all bets are off. Options: Majority Liberal gov't with Quebec returning to its traditional Liberal fold; resurrection of the BQ, because niqab's matter; majority/plurality CPC, because niqab's matter and Harper has a lot of money for a good ground game. Which happens? I dunno. But polling data says: watch Quebec!

      Delete
  16. Liberals continue to lead CPC both in the polls and the poll tracker.

    Harper is getting visibly desperate. Many are expecting elections fraud to rear its ugly head again as ABH really kicks in after Thanksgiving.

    ReplyDelete
    Replies
    1. But not the seat projection, which is all that matters.

      It does not matter which party gets more votes. Look at the US - in the last congressional election, the Republicans won a strong majority in the House, but the Democrats won millions more vote nationwide.

      Because this isn't a national race. This is 338 consecutive local races. It only matters who wins a majority of those.

      Given the vote distribution of the parties, I think it's possible not only for the Conservatives to win the most seats while losing the popular vote, but I think they could even win a majority while losing the popular vote.

      It doesn't happen often, but it does happen. The most extreme example is probably the 1921 provincial election in Alberta, where the Liberals won 34.1% of the vote, compared to 28.9% for the United Farmers. But the UFA won more than twice as many seats, 38-15 (with 5 more seats going to the Labour and Conservative parties).

      There's no reason to care about which party gets the most votes nationally.

      Delete
    2. Wait, ignore that UFA example. I didn't realise how messed up Alberta's election rules were back then.

      Urban voters get to vote 5 times? What?

      Delete
    3. I try not to engage trolls, but your reference to the CPC engaging in election fraud is over the top. This site should be about serious discussions of polling methods not about sleazy innuendo.

      Delete
    4. I don't think Harper is that desperate. He wants to win badly, sure, but I think it's more about trying to win a Majority. He looks fairly confident to me. I think he knows that he is back in Minority territory and will probably finish 1st in seat count.

      Delete
    5. Elections fraud is over the top for the CPC? The CPC were charged with In and Out elections fraud, misleading voters, elections overspending and oh ya there is Dean Del Mastro. Have the CPc turned over a new leaf? I doubt it.

      As for seat counts, the polls and seat counts both show Harper has no chance of a majority. This will end his career soon after the 19th.

      Delete
    6. Jimmythedeke has a point. The CPC in the past had no qualms doing things that were beyond the pale in terms of electoral fraud. Will they do the same again this time around, or should we assume that they have now reformed and are holding themselves to a higher standard? History suggests that corrupt organizations remain corrupt unless there is major change across the board.

      Delete
    7. The CPC was charged. Did anything come of that? Or were the accusations groundless?

      I insist that there's no such thing as the spirit of a rule. In-Out was an attempt to follow the rules in an advantageous way.

      When I worked for the Reform Party, I proposed a system whereby the party could generate revenue purely by siphoning money from Revenue Canada. The rules were written such that the party and a paper donor could both turn a profit without breaking any rules.

      As far as I know, the party didn't do it, but my understanding is that the Bloc did.

      Delete
    8. The Conservatives' new legislation allowing candidates to recomend people for poll clerk positions opens up a can of worms. Just one example: imagine that you are unemployed or underemployed, and you feel strongly enough about the A Party to volunteer for them, and in return the candidate has gotten you a day's work as a poll clerk. During election day, two people come in, a couple of hours apart, and don't volunteer proper identification. From your volunteer work canvassing, you recognize one of them as someone who hates Party A, and the other as someone who loves Party A. Are you going to be equally helpful to both of them in finding proper identification? Maybe, but you're definitely in a conflict of interest.

      The law that poll clerks MUST provide information to local party headquarters about who has voted is also problematic, especially when you consider that the polling clerk who is going to be repeatedly on the phone to report information to the various local party headquarters during Election Day was quite possibly canvassing for one of those parties the day before.

      Then there's the contradiction between the law that says you must be a Canadian citizen to vote and the voter ID rules which quite clearly specify that you don't need to provide the slightest bit of evidence that you're a Canadian citizen in order to vote.

      Delete
    9. Michael Sona was found guilty of one violation of the Elections Act (for preventing or trying to prevent a voter from casting a ballot, which carries a maximum penalty of a $5,000 fine and five years in prison) last year, while Andrew Prescott was granted immunity for being a witness. The trial judge concluded that it was unlikely that Sona acted alone in the commission of the offence. Sona received 9 months of imprisonment and 1 year of probation.



      The Etobicoke Centre results went to the Supreme Court and a by-election was overturned 4-3.

      Delete
    10. Eric,
      Comments insinuating election fraud are not appropriate in my opinion. I know you can't police everyone but when Jimmythedeke infers the Conservatives will use fraud to win I think one must draw the line and remove such comments.

      Delete
    11. To quote from Judge Mosley who looked at the issue of electoral fraud in the 2011 election:

      “I find that electoral fraud occurred during the 41st General Election... ”

      “In reaching this conclusion, I make no finding that the Conservative Party of Canada or any CPC candidates or RMG and RackNine Inc., were directly involved in any campaign to mislead voters,” he ruled.

      In other words, the CPC did not engage in electoral fraud.

      Delete
    12. No, there was not sufficient evidence to say the CPC was DIRECTLY involved, just the deputy campaign manager and a junior staffer accessed the Conservative voters' database CIMS and using this information collected due to CPC voter identification calls then targeted non-Conservative voters. And at least part of it occurred in Burke's campaign headquarters.

      Voter suppression is a key Conservative election strategy. The latest example is the "Fair" Elections Act. The CPC continues to abuse the system to the fullest extent possible and only occasionally crossing the line into indictable offences, which are systematically the work of individual members of the CPC and never the CPC organizationally.

      Delete
  17. Hi Eric,

    Long time fan, first time commentator.

    I was wondering how you factored in the political biases of certain pollsters. It is becoming clear to me with each poll they conduct that Mainstreet is a highly biased and partisan pollster that greatly inflates Conservative numbers. Their polls always look like outliners compared to other more reputable pollsters like Nanos and Forum. Remember these were the same guys who had that outliner poll back before the election that had the Conservatives at 38% when no other pollster did. Their results don't really seem to mesh with what other pollsters say, especially for Ontario.

    ReplyDelete
    Replies
    1. Based on their historical track record.

      If one pollster consistently inflates a party's numbers relative to the other pollster's results, that adjustment is added.

      For example, Forum consistently inflates Liberal numbers (of all the polls since the 2011 election, the top 14 Liberal results are all Forum polls), so their Liberal values get reduced a bit.

      Delete
    2. Eric does weight by past poll accuracy IIRC.

      Delete
    3. Because you think Quito Maggi is a Conservative? hahahaha NO Quito Maggi is a Liberal riding president and worked on Iggy's campaign. Definitely not a Conservative. The methodology may be weak but its not because they are partisan.

      Delete
    4. Ira - I do not apply house effects to poll results.

      Delete
    5. Quito Maggi has declared that he has worked for both Liberal & Conservative campaigns in the past.

      Bruce Anderson of Abacus has acknowledged that he has done work for Liberals & Conservatives.

      And Darrell Bricker & John Wright of Ipsos are known to be Conservatives. Bricker worked for the PC's back in the Mulroney days, I believe.

      Question is whether any of this matters or influences the polling.

      Delete
    6. To monkey cheese, i dont see how you see Mainstreet as an outlier when everyone but Nanos is in agreement within the MOE.

      Delete
    7. @darren abrey:

      I'm not sure what you're getting at, but Nanos is hardly an outliner on this one. Leger, Innovative, and Ispos have all showed a Liberal lead. Abacus is now showing a tie between the Liberals and Tories while EKOS is now showing a tightening race between the Conservatives and Liberals. So yes, I do think that Mainstreet is an outliner due to their inflated Conservative numbers, flawed methodology, and possible biases.

      Thanks for the information Craig.

      Delete
    8. @Darren abrey, that's an interesting interpretation of "everyone" when Innovative/Hill Times had the CPC at 29% vs. 31% LPC on the same day. Ipsos Reid had them at 32/33 and 33/32 on the two polls bookending the Mainstreet timeframe. EKOS had them at 33.4% vs 26.7% with an MOE. Abacus Data was 32% to 29%.

      So in summary, "everyone" means just one Forum poll agrees with Mainstreet at the outer edge of the MOE.

      Delete
    9. Éric - Sorry, I thought you did.

      Delete
  18. Mainstreet polled 436 18-34 year olds and 2031 65 year olds. Only 1412 between 18-49. Old people vote more for sure but that sampling seems wonky. Doesn't appear to be weighted.

    ReplyDelete
    Replies
    1. No wonder why Mainstreet's polls always inflate Conservative numbers. They are underestimating the youth vote which is historically not very friendly to the Conservatives. A more reputable pollster would try to have a balance between all demographics.

      Delete
    2. If they didn't weight it then it's definetly off. Seniors vote more then younger people but not five time more. That being said I see no mention of weighting in their report... which doesn't mean that it wasn't weighted, just that we don't know. I suppose someone could run the numbers to figure it out based on what was provided but I'm not gonna volunteer.

      Delete
    3. Remember Éric's calculations of a few years ago. He found that a representative sample was more likely to match actual voting patterns if you assumed that everyone 55+ voted twice, and people under 30 didn't vote at all.

      Mainstreet might be trying to account for something like that.

      Delete
  19. With the average of Octobre 4th (for the ninthe week), my model gives:

    133 CPC (+6)
    102 LPC (+2)
    95 NDP (-12)
    7 BQ (+4)
    1 GPC (even)

    By region, it gives:

    Atlantic
    23 LPC (+1)
    6 CPC (+2)
    3 NDP (-3)

    Québec
    48 NDP (-6)
    14 LPC (even)
    9 CPC (+2)
    7 BQ (+4)

    Ontario
    55 CPC (even)
    46 LPC (+1)
    20 NDP (-1)

    Prairies
    16 CPC (+3)
    6 LPC (even)
    6 NDP (-3)

    Alberta
    30 CPC (+2)
    3 LPC (-1)
    1 NDP (-1)

    British Columbia
    17 CPC (-2)
    16 NDP (+2)
    8 LPC (even)
    1 GPC (even)

    Territories
    2 LPC (+1)
    1 NDP (even)
    0 CPC (-1)

    ReplyDelete
    Replies
    1. Which three ridings does the LPC get in Alberta in your model?

      Delete
    2. The LPC gets Calgary-Centre (easily), Calgary-Skyview (tight race) and... never mind the third. I made a change a while ago in my forumla that I haven't had the time to implement, and the result changes with it. So the LPC gets 2. The NDP wins Edmonton-Strathcona easily.

      Delete
  20. The Mainstreet poll was conducted before the last French debate. That makes it almost a week old. How can it take so long to release information which is very time sensitive?

    ReplyDelete
    Replies
    1. Valid point. Any poll at this late stage in the campaign that is more than a few days old risks being outdated very quickly.

      Delete
  21. Eric: Do you post accuracy ratings for the polsters, like 538 does ?? If so, where would we find them ?

    ReplyDelete
    Replies
    1. It is Nanos at 50 and the other 7 pollsters at 7 each.

      Nanos is the big outlier polling a huge Liberal lead. Without Nanos polls the discussion would be how close we are to a CPC majority and whether Trudeau / Mulcair would really not support the CPC party that has 40-50 more elected MPs than the runner-up.

      Delete
    2. BCVoR is a well known Harper Conservative partisan cheerleader around these parts, so anything he says should be taken with a grain of salt.

      As I mentioned earlier, Nanos is hardly an outliner on this one. Leger, Innovative, and Ispos have all showed a Liberal lead. Abacus is now showing a tie between the Liberals and Tories while EKOS is now showing a tightening race between the Conservatives and Liberals.

      Nanos has consistently been one of the most accurate pollsters in Canada. To dismiss Nanos is nothing short of partisan wishful thinking. Nanos should be commended for its daily polls.That must take a lot of time and resources.

      I don't think any party is heading for a majority this time around.

      Delete
    3. The Libs NDP will topple the CPC no matter how large their minority. Guaranteed. Even if they get 169 seats, which they won't..

      Delete
    4. Nanos does skewed with high Liberal numbers, but without them in the mix, the CPC doesn't have a sizeable lead (beyond a statistical tie).

      Using Oct 6 numbers and using the weightings dropping Nanos, CPC up 0.87% to 33.2%; NDP up 1.25% to 26.2%; LPC down 2.35% to 30.2%; BQ up 0.18% to 5.1%; Green up 0.21% to 4.9%; Other down 0.16% to 0.5%.

      Delete
    5. One major question about supporting a CPC minority is if they receive a plurality of votes. It's still possible that Harper wins on seats, but Trudeau wins on votes.

      Delete
    6. Abacus, Ipsos, Innovative Leger all outliers too? The only polls showing CPC high are Mainstreet and Ekos. Even Ekos' CPC numbers in latest poll have fallen.

      It seems well known outlier Mainstreet is the only poll that shows the CPC even any where close to a majority.

      Harper is in deep trouble. That is why he is doubling down on the niqab thing. Desperation.

      Delete
    7. If the CPC is tied in the polls with the Liberals going into election day, the CPC wins the election.

      The Liberals need to be ahead by several points to be confident they'll win a plurality.

      Delete
    8. 169 would be enough - they'd just need to get an opposition member to be speaker.

      Delete
    9. Harper is doubling down on the Niqab issue because they are now leading in Quebec, according to several polls without much notice form the CBC et al. That's huge.

      Delete
    10. You really, really can't see truth or reality can you ???
      That's a blatant lie shosh !!

      Delete
    11. If the CPC is tied in the polls with the Liberals going into election day, the CPC wins the election.
      ====================================
      Yup, and Harper and the CPC will lose the government shortly afterwards.

      Delete
    12. Jimmythedeke,

      Young Trudeau has already stated he will not work with Mulcair, until that changes the Conservatives will remain Government.

      It is the NDP who are in very serious trouble, they will lose some big names on the 19th unless they can somehow reverse the trend; Deputy leader Megan Leslie looks likely to be defeated, former Nova Scotian Opposition leader Robert Chisholm looks likely to be defeated, Peggy Nash look likely to go down in defeat as does Matthew Kellway in Beaches-East York, and Murray Rankin is also in trouble in Victoria, which should be the NDP's safest riding. Never mind that at this point they look set to lose up to half their Quebec caucus. It is fairly certain the NDP will win less than 100 seats, the only question is how low will they go?

      Delete
    13. "Young Trudeau has already stated he will not work with Mulcair, until that changes the Conservatives will remain Government. "

      And those of us who are older know that such statements made prior to an election outcome are nonsense. If Harper has a plurality, but well shy of a majority, and LPC+NDP is well above 170, I have absolutely no doubt they will come to some agreement.

      Delete
    14. "That's a blatant lie shosh !!"

      I wondered about it when I read it; it isn't a lie; the October 6th EKOS poll has the CPC at 28% in Quebec, ahead of the LPC at 24.5% and the NDP at 23.8%. The sample size, 337, is such that the MOE is 5.3% and that is (like all polls) with 95% confidence, which is rather low given the vast number of polls (well over 20 ...) this election. It isn't however, a lie by her.

      Delete
    15. The collapse of the NDP in Quebec creates signficant opportunity for the Bloc. I can imagine a scenario in which the Bloc wins 6 seats and holds the balance of power.

      However, the growing strength of the CPC in Quebec limits the Bloc's potential.

      Delete
  22. It would be useful to adjust for periodicity. This is particularly noticeable for Nanos NDP vote numbers in Quebec. They do worse on weekends and bounce back on work days. Not sure whether they change their mind depending who they are talking to or whether they reach a different population on weekends.

    ReplyDelete
    Replies
    1. The numbers in Quebec are going to be difficult for all pollsters to figure out until it's clear what is going on there.

      It looks like the Conservatives & Liberals (& BQ) have both gained at NDP expense. EKOS shows Conservative & Liberal gains in QC today, and NDP losses, so EKOS agrees with Nanos on QC drop for the NDP, as did Ipsos last night.

      Delete
  23. There are a lot of good riding polls out lately and that's great news for the anybody but Harper camp. There seems to be a website that's compiling the individual riding polls to give people guidance:

    http://www.votetogether.ca/riding/list/

    That should keep the ABH crowd happy.

    ReplyDelete
  24. Interesting that the QC NDP are taking the niqab hit while the LPC is seemingly getting a free pass.

    Perhaps the LPC is beginning to get conferred the mantle of the party that can win against Harper and are winning over 2nd choice voters.

    The CPC needs to equalise ON NDP and LPC votes. That was working for a while, but the LPC is pulling ahead of the NDP in ON.

    Away from 308, we have to ask how many voters are aware of the state of play in their own riding.

    Two examples out of many:

    Lethbridge: LPC vote effectively a CPC vote

    Eglinton Lawrence: NDP vote effectively a CPC vote.

    ReplyDelete
    Replies
    1. Last year about this time, it looked like the LPC was on track to win as it does now. Then we had the lone wolf nutbar attacks. The CPC went on top early this year, then an NDP spring.

      Whichever party gets the break, if any, in ON will likely come out on top.

      Delete
    2. I think Quebeckers who were willing to vote Liberal before are Quebeckers who don't care about the niqab issue (or oppose Harper's position on it), so they;re still willing to vote Liberal now.

      I'd love to see a current linguistic breakdown of Québec. How well are the Liberals doing with Francophones?

      Delete
    3. Not that interesting... they have different pools of voters in Quebec do they not? Urban v. Suburban/Rural, more federalist v. more nationalist. So the niqab debate probably hits the NDP more close to home.

      Just a guess.

      Delete
    4. People should vote for who they think best. Strategic voting is nonsense, you replace your preferred candidate with one who doesn't hold your views. The NDP likes to use "strategic voting" since, they are not Canadians' preferred political party.

      Delete
    5. That's a pretty simple case of them having different voters. At least per the polls, most Canadians seem to be opposed to people wearing niqabs in these ceremonies, but it's primarily French-speaking Quebeckers for whom it's an issue they actually vote on. The Liberal base in Quebec is primarily Anglo- and Allophone.

      Delete
  25. So we see with the abacus poll the behavour Eric talked about on CBC yesterday. Online v. IVR v. Live Agent.

    Interesting. If the split continues onto election day in addition to being a referendum on the future of the country it'll also end up as a referendum on IVR vs. Live Agent.

    ReplyDelete
    Replies
    1. The main significance of the Abacus poll seems to be that it confirms the NDP drop that Nanos picked up on.

      NDP has now fallen quite far behind the Conservatives & Liberals in most polls.

      Delete
  26. Another factor; are conservatives more likely to answer a poll, than the other two parties.

    A lot of conservatives, are older and have free time on their hands, and are entrenched in their ways. They have no problem telling a pollster that they are strong conservatives. Meanwhile, the younger generation is more likely to hang up the phone. And some people even lie on purpose as they want to mess up the polls. There was a study, and I think that number is around 10%

    I the UK all the polls were way off, and the reason was that people was too embarrass to say they would vote for Cameron. Put when they put there mark on the ballot in private it was for Camron. I just pray that does not happen here.

    ReplyDelete
  27. In my own riding in BC the niqab issue scores because of the demographic - the "good ol' boy" who probably never "liked" foreigners, but was raised not to say so. They might have been NDP labor side, but flip when the bigotry is validated by others.

    Maybe the LPC gets free pass because they don't really have the demographic that this works on to begin with.

    ReplyDelete
    Replies
    1. "In my own riding in BC the niqab issue scores because of the demographic - the "good ol' boy" who probably never "liked" foreigners, but was raised not to say so. "

      I don't think it is that simple at all. There are many, many immigrants who are uncomfortable with the niqab and burka, not just "old-stock Canadians," though they often cannot quantify why. Here's a couple of points people miss: choosing to do something is not, in itself, sufficient to make it acceptable. Example, you cannot choose to enslave yourself. Further, the requirement that a female wearing a niqab can only have her identity visually verified by a female constitutes discrimination against males. Why would we, as Canadians, override our own Constitution that explicitly prohibits discrimination on the basis of sex, so as to satisfy someone who is not a Canadian? Since when does a clothing choice by someone who is not Canadian, and said choice is not required by any religion, override gender equality?

      While we're at it, would you be OK with a person wearing a niqab providing services to you? If so, can I presume you would also be OK with a person wearing KKK gear, since it has at least as strong, if not stronger, claim as "religious" clothing?

      Delete
    2. You are conflating the Rights of Citizens/Permanent Residence with the Rights of the State. There are requirements that a female officer searches female passengers in airport, and this would just be an extension of that.

      First and second generation Canadians represented 39.3% of the population in 2011. Meaning, as Canadians, it's highly likely that how we treat "non-Canadians" is how we or our parents would have been treated.

      FYI, we have the fundamental Right to freedom of conscience and religion, but there is no such explicit Right for equality, gender or otherwise.

      The KKK robes are not religious garments, or are you suggesting that lynchings and burning crosses are a form of Christian worship?

      Delete
  28. The simple average of the 8 pollsters most recent published poll :

    CPC 32.9 , Liberal 31.0, NDP 25.2, Green 5.4, BQ 4.9

    Then weighted by sample size:

    CPC 33.7 Liberal 30.5, NDP 25.0 Green 5.8 BQ 4.7

    ReplyDelete
  29. The cyclical weekly fluctuations bode well for the Conservatives, as they tend to do well on Mondays, and the election is on a Monday.

    ReplyDelete
    Replies
    1. Ira, I believe you are confusing how the Conservatives do on the weekend and Monday. Looking at an average of all polls that contain a given day of the week:

      CPC: best on Sun, Sat, Fri, Tues, Mon, Thurs, Wed.
      NDP: best on Wed, Thurs, Tue, Fri, Mon, Sun, Sat.
      LPC best on Fri, Thurs, Sat, Sun, Wed, Tues, Mon.
      BQ best on Sat/Sun, Mon, Tue, Fri, Wed, Thurs
      GP best on Mon, Wed, Tues, Thurs, Fri, Sat/Sun.

      Delete
    2. This is funny. Thanks for the light hearted satire. As is the day of the week will move votes


      If the wind is from the east it will also favour the CPC. LOL..

      Delete
    3. "If the wind is from the east it will also favour the CPC. LOL.."

      You may consider it funny, but weather demonstrably affects political outcomes in elections. There is a large research literature in the area and, to quickly summarize it, bad weather reduces voting, but does so differentially in voting groups. E.g., in the US, bad weather on election day is good for the Republicans.

      Delete
    4. Well yes...it seems a bit odd that day of the week polling would make any difference, but like the people posting about it...I too also starting thinking is there a trend going on with the cons up mid week and the libs up on weekends?....it sure does look like it..but that is more likely to be related to who is being polled on weekdays vs weekends.

      Delete
    5. Why are you so confident it doesn't, Jimmy?

      Why would you ever hold an opinion in the absence of supporting evidence?

      Delete
    6. Polling data is often skewed by short-term occurrences (for example, sports fans watching NFL), which is why polls are conducted over longer periods (often a week or two) to balance it out.

      Another random stat, of the previous 5 elections in 15 years, the two Sunday elections had an average of 3.4% lower turnout and both resulted in minorities.

      Delete
    7. East wind is straight out of Mordor - of course it benefits Harper :-)

      Delete
  30. Sorry Ira and BC but the more I talk to people on the street the more prominent the ABC is !! Harper will LOSE !!!

    ReplyDelete
    Replies
    1. Sorry Peter, The polls and seat projection disagree with you! I live in a Liberal riding and went to the candidate debate, the loudest cheers were for the Tory.

      All this talk, especially referencing Harper, smells of desperation on the part of the Liberals and particularly the NDP. The evidence demonstrates the Grits and NDP can't win so their only hope is to talk up their own very slim prospects by inventing fictitious movements. Harper is not guaranteed re-election but the polls, seat projections, and the mood on the street all favour the Conservatives.

      Delete
    2. Peter: Premier Dix would suggest you widen your circle of friends.

      Even Eric's CBC heavily Nanos based seat projection has Harper with 132 seats with only 13 CPC seats in BC..

      We both fully understand that the CPC has at least 15 safe seats in BC.

      Delete
    3. Harper will win his seat. The CPC will likely win the most seats. And who really cares about what people on the street say to you. I could do the same thing in Calgary and based on that forecast a massive CPC majority.

      Delete
    4. Harper cannot be written off. Particularly since he is still *leading* in several of the polls and may have an additional advantage of a couple of points on Election Day in terms of turnout, hidden CPC vote of people who don't want to admit they are voting Conservative.

      I think the Cons have to be behind by at least a few points going into October 19th in order for the Liberals to beat them.

      More NDPers will have to move over to the Liberals in order for that to happen. There does appear to be some movement from the NDP to the Liberals right now, but it has not reached a point of critical mass.

      Delete
    5. "The CPC will likely win the most seats."

      They may; what happens next, though, will likely be a failed attempt to form gov't by Harper, a request for a new election by him is possible, but would almost certainly be denied. Then you get a new PM, either Mulcair or Trudeau, depending on who came in second. How long that lasts would depend on the various negotiations between Mulcair, Trudeau and Harper.

      Delete
    6. I don't know about that BCVoR. BC can be mercurial, and the realignment of ridings has mixed things up a bit.

      It does seem likely that the CPC will win the most seats in this election. Will they win enough to maintain power? There's really no way to say.

      The people who insist Harper will win and the people who insist Harper will lose are both making baseless claims.

      Delete
    7. Ira: When I wrote that "Harper will win" I meant exactly that. Harper will almost assuredly win his seat in Calgary. Harper winning his seat doesn't mean that the CPC will win the most seats.

      Delete
    8. I understood what you meant, Lord Feltham. My response was perhaps drafted inartfully, if it gave you the impression I did not.

      Harper will win his seat easily (not as easily as Jason Kenny will next door).

      Delete
  31. Harper is done. No majority = career ender and lose of government. Things look really bad for Harper.

    ReplyDelete
    Replies
    1. He's been Prime Minister for nearly 10 years. He's the 6th longest serving Prime Minister in Canadian history. It's hardly a personal tragedy for him if he stops being Prime Minister in a few weeks.

      Delete
    2. Not winning a majority isn't a career ender. From 1956 to 1967, Diefenbaker and Pearson had three four consecutive head-to-head elections (PC majority, PC minority, 2x LPC minority). Since then, only if a party didn't form a minority government did a leader step down.

      Delete
  32. If the CPC picks up another 5%, there's a possibility the ABC turnout will get an extra dose of motivation.

    ReplyDelete
  33. What I find strange, is by looking at the long term poll averages, the Conservatives and Liberals have flatline over the past few weeks, while the NDP has crashed. No party has taken more than 1 or 2% of those votes. Where is the missing 8 to 10%, of former NDP votes?

    ReplyDelete
    Replies
    1. "Where is the missing 8 to 10%, of former NDP votes?"

      We long ago gave up on the Liberals and CPC. We liked the NDP when they looked more centrist. They don't look so good now, and so we're in a mess. I, for one, will likely vote for a fifth-party choice, since the Greens seem quite wonky this time around.

      Delete
    2. I expect to vote Libertarian.

      Delete
    3. Most will vote ABH. Many will now start voting Liberal as the most likely to defeat Harper.

      Delete
  34. The partisanship in these comments is always good for a laugh, especially by people like BC Voice of "Reason" and Formerly Carbonear Pete. It's amazing how far people go to try and discredit the work Eric does here to suit their own partisan agenda. Dismissing credible pollsters like Nanos is nothing short of uber partisan wishful thinking.

    Interesting results from Forum today. Liberals out in the lead nationally, but also leading in the Atlantic, Ontario, BC (although BC remains a true threeway tie), and the Prairies. NDP leads Quebec (Liberals 10 points behind). Alberta is the only place where the Conservatives lead.

    http://poll.forumresearch.com/data/545503eb-1598-434e-8d31-a72f68b05e27Federal%20Horserace%20News%20Release%20(2015%2010%2007)%20Forum%20Research.pdf

    This is a contrast from Forum's last poll which had the Conservatives leading by almost 10 points, but seems to correspond with the trend of a tightening two-way race between the Liberals and NDP. Hence why I think there will be no majority this time around.

    ReplyDelete
    Replies
    1. Whoops, that should say tightening two-way race between the Liberals and Conservatives.My bad.

      Delete
    2. "seems to correspond with the trend of a tightening two-way race between the Liberals and NDP."

      I agree that there are partisan commenters, but would not limit it to the conservative supporters. Yes, BCVoR and FCP are clearly CPC partisans, but equally Jimmythedeke and Peter Meldrun are anti-Harper partisans, going so far as to claim that only fraud can help Harper now. I'm curious as to why you would present just one side of the partisans.

      "seems to correspond with the trend of a tightening two-way race between the Liberals and NDP."

      I'm not quite sure what you mean by this, given that the national picture shows a tight race between the Liberals and CPC, while the NDP have unambiguously crashed according to all polling firms, being somewhere between 5 and 11 percentage points behind the send-place alternative.

      Delete
    3. Those 2 specifically are not the only ones...but...I am here to read Eric's analysis not theirs. The thing to remember is that people with strong partisan blinders will always appear to a blog like this...

      Delete
    4. Interestingly the only partisans I seem to see on this site are pro-Liberal and pro-Conservative (and uniformly anti-NDP). Maybe it's an issue of more enthusiasm on the part of the Grits and Tories that they are up in the polls, but if anybody looked on here you would be hard-pressed to think it's actually a relatively three-way race in the election.

      Delete
  35. Conservatives drop from 6 point lead to trail Libs by 4 points in one week in latest Forum poll. Thats a 10 point shift. ABH is now building steam. Harper is in deep trouble.

    ReplyDelete
    Replies
    1. How could ABH possibly account for a drop in Harper support? If that gap closed simply by having the Liberals catch up, sure, but the ABH vote was never voting for Harper (so deep is their supposed hatred for him), so how does ABH growth explain a CPC drop?

      Do you think it's turning a bunch of non-voters into voters?

      Delete
    2. Its both. ABH and Harper losing support and popularity. His time is up. Canadians are tired of his petty, vindictive, dishonest government. The Liberals are going to surge over the next week and the CPC are going to slip further into second place.''

      Harper is done.

      Delete
  36. Eric, I wonder what your thoughts are re strategic voting historically vs new campaigns like votetogether.ca? Does the technology/connectivity make this a viable movement? Or is the number of voters interested still too small? If you think their efforts might make a difference, how might results differ from predictions? Are strategic voters simply captured as 'undecideds' right now or are they hidden?

    ReplyDelete
    Replies
    1. I think there are a lot of strategic voters, but that most of them aren't aware of these sorts of initiatives.

      I think we're seeing them move in the polls, so it might not have too much of an impact on the projections.

      Delete
    2. I think strategic voters are minor next to the much larger subset of voters who just support who they think is the most likely to win (and to whom platform and policies are secondary to making their vote count). There seems to be a big enough bandwagon effect in elections that what might otherwise be a minor dip in a party's polling position gets magnified by all the bandwagon jumpers who hop to whoever is being talked of as having the momentum.

      Delete
  37. Éric,

    I'm curious as to how you're handling the new EKOS daily rolling average polls in the Poll Tracker, in particular given the seemingly inconsistent number of daily interviews, as opposed to Nanos who conducts exactly 400 interviews each day. This could help us decide how best to graph the EKOS polls over at: https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2015

    Thanks,

    Dom

    ReplyDelete
    Replies
    1. I'm handling them the same way as Nanos, reducing each old poll that overlaps with a new one to 33%. That takes care of the date weighting. The sample size weighting takes care of the differing sample sizes.

      Delete
  38. Eric, I'm wondering how you incorporate individual riding polls

    It is clear that they have some impact on the riding projections. We saw both Peterborough-Kawartha and Pontiac change from Tory to Liberal projected wins after riding polls indicated a Liberal lead.

    Also, are riding polls included somehow in the regional and national poll average numbers?

    And a final question - has anyone polled support for Forces et democratie in Quebec? It looks like the "other" vote is particularly high there, and I'm wondering if it corresponds to this new party. The riding projection for Avignon-La Mitis-Matane-Matepedia also looks as though they may keep that seat. Am I reading that correctly?

    ReplyDelete
    Replies
    1. It is explained in the methodology page here:

      http://www.threehundredeight.com/p/forecasting-methodology.html

      ...about three-quarters of the way down.

      As to FeD, I haven't seen a single national, regional, or riding poll that mentions them. But you're right, I give Fortin a chance in his riding.

      Delete
  39. It is hard to believe the latest Forum poll or at least parts of it. There is no way that the Liberals are leading in the Prairies with 40% popular support.

    ReplyDelete
    Replies
    1. Don't forget the MOE on regional numbers are very high. For 193 sample size in 2.427M people, it's 7.1%, so it's a statistical tie between 40% and 35%.

      Delete
    2. CPC is even losing seats in Alberta.

      Delete
  40. Yeah Forum thinks the CPC are at 35% in the Prairies. Um are they smokin' crack?

    The CPC are at 50% in the "Prairies", 60% in Alberta, 50% in Saskatchewan, 40% in Manitoba or thereabouts.

    ReplyDelete
  41. Ekos... http://www.ekospolitics.com/index.php/2015/10/liberals-now-have-slight-lead/

    ReplyDelete
  42. Forum Leger, Ekos, and Nanos all today agree Liberals are in first and rising. Harper is in big trouble and slipping.

    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.