Friday, June 29, 2012

Majority support for federal NDP in Newfoundland & Labrador?

Environics released a new poll earlier this week, showing that the New Democrats continue to hold a narrow lead over the Conservatives, mostly unchanged from their last poll of May 7-9. But a sub-regional breakdown provided to ThreeHundredEight shows that the NDP is running away with things in Newfoundland and Labrador - and struggling in Saskatchewan.
Compared to Environics' last poll, the NDP is down one point to 35% while the Tories are up one to 33%, an insignificant shift even on a large sample size such as this. The Liberals are unchanged at 19%, while the Bloc stands at 7% and the Greens are down one to 6%.

The Conservatives are up four points in Ontario to 40%, while the NDP is down two to 30%. The gap stands at 10 points, but that has been bouncing around from poll to poll. The Liberals, down three to 23%, have been struggling in most surveys.

In Quebec, the New Democrats are down one point to 44% and the Bloc Québécois is up one to 26%. The Liberals are up two to 15% and the Tories are down three to 12%, none of these being very significant changes in support.

The gap has narrowed in British Columbia, where the NDP is down six points to 37% and the Tories are up five to 34%. The Liberals are well behind the pack with 21% (+1), while the Greens are down two points to 8%.

Alberta and the Prairies are standard fare, while the NDP is up 13 points in Atlantic Canada to 44%. The Liberals are up 10 to 30%, while the Conservatives are down 15 to 23%. These wild swings are likely statistical noise, but the sample is more robust than usual: Environics over-sampled the region to get useable Newfoundland numbers (but of course, those were scaled back to represent an accurate weight within the region).

None of these regional results are remarkable, they generally line-up with what most other polls show to be the case. But the sub-regional results are fascinating, giving us a look at the country's three largest cities (and their metropolitan areas), how the vote breaks down in the Prairies, and where things stand on the Rock.
We shall start out West, where the New Democrats lead in Vancouver with 41%, putting them ahead of the Conservatives by eight points. The Liberals stand third with 24% support in and around the city.

Moving to Saskatchewan, we see that the NDP is not doing as well in the province as has been expected. Most of their gains in the Prairies appear to be heavily weighted to Manitoba, where the race is quite close.

In Saskatchewan, the Conservatives are down nine points from the election to 47% support, but the New Democrats have slipped four points to 28%. This does not put them in range of winning any seats, maintaining the status quo of 13 Conservatives and one Liberal (they are up four points to 13% in the region). The Green result, at 12%, is probably too high, but it is difficult to know for sure where those votes might go.

The Conservatives have dropped 15 points in Manitoba to 39%, while the NDP is up nine points to 35%. The Liberals, meanwhile, are steady with 17% support. This would likely deliver four seats to the NDP and one to the Liberals in the province, with the remaining nine going to the Tories. That is down two seats for them.

In Toronto, the race is a close one: the Conservatives have 33% support, the NDP has 31%, and the Liberals have 28%. This suggests that the NDP is starting to breakthrough into the outlying suburbs of Toronto, but that the Liberals are still a force in the city.

The NDP is running away with it in Montreal, where they have 52% support. The Bloc is the runner-up with 21%, putting them narrowly ahead of the Liberals. They have 16% support, much of that undoubtedly concentrated on the West Island. The Tories are completely out of the race with 8% support in the area.

And that brings us to Newfoundland and Labrador. The sample for the province isn't huge, with an MOE of just under eight points, but it is large enough to give us an idea of what is going on. I'm also told that a larger sample has backed up these results. Environics puts NDP support in Newfoundland and Labrador at 59% support, up an amazing 26 points since the election. The Liberals are down 16 points to 22%, while the Tories are down 10 points to 18%. With this sort of swing, the New Democrats could win four seats in the province, with the Liberals barely holding on to three.

On these numbers (excluding the city and NL results, however), the Conservatives win 142 seats to 120 for the NDP and 40 for the Liberals. The Bloc would win five seats and the Greens would retain one.

The Tories win 17 seats in British Columbia, 27 in Alberta, 22 in the Prairies, 70 in Ontario, three in Quebec, two in Atlantic Canada, and one in the North. As usual, it is the combination of the West and Ontario that keeps the Conservatives on top.

The NDP wins 13 seats in British Columbia, one in Alberta, four in the Prairies, 25 in Ontario, 60 in Quebec, 16 in Atlantic Canada, and one in the North.

The Liberals win five seats in B.C., two in the Prairies, 11 in Ontario, seven in Quebec, 14 in Atlantic Canada, and one in the North.

As mentioned, the top-line and regional numbers add to the general picture that is forming about where the parties stand throughout the country. But the sub-regional ones paint a slightly different landscape. While caution must be used with the small samples, they suggest that the NDP is not making the inroads it needs to make in Saskatchewan and that the party is poised to do exceptionally well in Newfoundland and Labrador. The Conservatives are still in play in Vancouver and Toronto, but are being challenged more and more in the Ontario capital.

Environics gives us some interesting numbers to mull over the Canada Day weekend.

72 comments:

  1. South Parkdale Jack29 June, 2012 16:56

    Absolutely shocking !!! What a nice Canada Day present. Cheers to you all, even BCVofR.

    ReplyDelete
    Replies
    1. It's not that shocking. NDP have been on the rise in Newfoundland for quite a while now, it was only a matter of time before they get majority support in that province.

      Delete
    2. South Parkdale Jack29 June, 2012 21:29

      It was said in jest. I think it's GREAT. And I always thought the NDP would break through in Quebec, just a little delay with the emergence of the BLOC. Cheers again.

      Delete
  2. Quebec may lose the title to Newfoundland of being the most "orange" province in Canada!

    ReplyDelete
  3. With only 1694 respondents nation wide, the Newfoundland and Labrador sample size is 25 or 26 - which is a +- 19.6. The sample in Saskatchewan would be about 50 which is +- 13.6. These sample sizes are completely statistically irrelevant because they are not so much a bell curves but an almost flat line.

    I know of no one that would take a sample size of 50 as being an accurate reflection of public opinion.

    The results for Vancouver, without knowing how they defined Vancouver, in a best case scenario has a huge statistical margin of error. In fact, other than Ontario and Quebec, the sample size for all the other break outs are too small to have any validity.

    ReplyDelete
    Replies
    1. Or, you know, I actually have the sample sizes here in front of me: 153 in Newfoundland and Labrador, 86 in Saskatchewan.

      I do know how the cities are defined (as I spelled out in my post), and the sample sizes are 96 for Vancouver, 167 for Toronto, and 189 for Montreal.

      But thanks for thinking I'm that daft.

      Delete
    2. And let me point you to the text from my post:

      "These wild swings are likely statistical noise, but the sample is more robust than usual: Environics over-sampled the region to get useable Newfoundland numbers (but of course, those were scaled back to represent an accurate weight within the region)."

      and

      "And that brings us to Newfoundland and Labrador. The sample for the province isn't huge, with an MOE of just under eight points, but it is large enough to give us an idea of what is going on. I'm also told that a larger sample has backed up these results."

      Delete
    3. I do not think anyone should ever be reporting on any polling numbers as being legitimate on sample sizes of much less than 500. The way you show the numbers gives them a legitimacy they do not deserve.

      Meanwhile their whole nation wide poll numbers are wrong because they have to significantly scale the numbers for Newfoundland etc and therefore they can not claim to have a poll that is 1694 respondents and use that to set the margin of error.

      The very fact Environics reported such small samples and combined vastly over sampled areas with under sampled says to me they do not understand the math behind the statistics and methodolgy of polling.

      Delete
    4. As any qualified statistician will confirm, a sample size under 300 is basically "useless" in terms of coming to a somewhat accurate political sounding.

      Just sayin'.

      Delete
    5. Hmm, yes, I suppose that means all those regional samples of 300 or less showing the same general results are just coincidental. You're right, the Greens could be at 80% in Alberta - WE JUST DON'T KNOW.

      Delete
    6. I should also add that you do not say in your post what the areas are that are covered by Vancouver - you say metro, but that can mean different things and federally the ridings do not follow the boundaries of Metro Vancouver.

      A sample of 96 in Vancouver is a meaningless tiny sample and had no validity as a measure of public opinion.

      What do you think is a valid sample size and how much messing with the raw stats by the pollster is acceptable to you?

      Delete
    7. They are not data you can rely on statistically. By dead reckoning you can get to the same numbers and that does not make that any more accurate methodically.

      Also the variation in the small regional samples is so large that I disagree that they are close. Close is within 1 percentage point, everything else has missed

      Samples between 300 and 500 are of limited benefit because the curve is too flat. Samples of over 500 are the only starting point of having reliable data

      Delete
    8. Well, since you have a major problem with how every polling company reports their data, I suggest you take it up with them. You're only interested in giving a lecture, anyway.

      Delete
    9. I have been trying to get things improved and I wish the media and public would better understand how badly most of the polling is done in Canada.

      You need to take a much more critical look at how the polls are conducted and how the statistics work within them. You do not do enough of that and you report headline numbers from small samples as if they have some connection to relevance. That is not fair to anyone.

      Delete
    10. I do not do that willy-nilly, I only report on significant regional shifts if they appear to be potentially statistically significant.

      The MOE on the NDP's support in NL on a sample size of 153 is +/- 7.8%, which means that, yes, this poll suggests they have majority support. I was also told that a more robust sample had the same thing.

      And the numbers do have relevance within the context of other polls that suggest the same thing. One poll here and one poll there is very different than tracking a half-dozen to a dozen polls every month.

      Delete
    11. I am not convinced that with 153 responses you can get a representative sample based on who will be voting in the election. One of the problems is that the pollsters do not know who voted in the last election and who will vote in the next election. Without knowing that the poll has trouble being representative at all.

      If there is any weighting of the 153 respondents the margin of error rises very quickly because you have to calculate it based on a smaller sample because you are saying the 153 is not an accurate reflection of the population.

      For a regional shift to be significant it needs to a large enough sample done at least three times by the same pollster. The problem is that no one is doing this.

      Our problem in Canada is that because there are so few public political opinion polls everyone digs into each one as if it is valid as every other one. It is also why I only comment on polls some of the time, there are very few I see out there that I have confidence that they are done in a statistically valid way - very few pollsters publish enough information in their press releases that I can tell what they did and how they did it. The way Environics did this poll says to me that there is no statistically valid data we can take away from it.

      Delete
    12. "Our problem in Canada is that because there are so few public political opinion polls everyone digs into each one as if it is valid as every other one."

      I think you'll be hard-pressed to find me saying that the result of any one poll is absolute, unassailable truth.

      If you want me to write every day that "this could mean something, but probably means nothing at all", I think you are erring on the side of caution.

      Delete
    13. Also, I believe you are being unduly harsh on Environics - especially considering you are working from a very small amount of information. The amount of weighting was not extremely high.

      Delete
  4. Somehow I would expect the NDP to do better than 4 seats, and more like 6 with those results. They are already quite saturated in St. John's East (winning 71% there) and somewhat in St. John's South-Mount Pearl (earning 47% of the vote last time around). I somehow doubt that the NDP will end up winning 98% of the vote in St. John's East and 75% in South. Plus, I think that a lot of this has to do with Harper's changes to the unemployment benefits, something that I would also presume would have a stronger impact outside in Avalon than inside it.

    Anyhow, just some speculation.

    ReplyDelete
    Replies
    1. The model puts the NDP within three points in Avalon, 11 in Bonavista, and five in Humber.

      Remember, the Liberals are dropping by a factor of .58, which means in Bonavista and Humber, where they had 58% and 57% support, respectively, pushes them down to about 34% and 33%. The NDP is up by a factor of 1.81, which only pushes them from 14% and 16% to 25% and 29% in those two ridings. Smooth out those numbers (and the others) to get to 100%, and you get my projected results.

      It might be counter-intuitive, but the Tories still did not win every seat despite getting almost 70% of the vote in Alberta either.

      Delete
    2. Right, but wouldn't that factor put the NDP over 100% in St. John's East? A smaller rise in the St. Johns ridings could put the NDP over the top in Avalon and perhaps Humber - not to mention the fact that Labrador is a much smaller seat than the rest, and thus the proportional gain may be higher.

      Delete
    3. No, the model ensures that everything adds up to 100%. So, for example, when the parties add up to over 100% all the parties are reduced proportionately so that it will add up to 100%.

      Similarly, when the parties add up to less than 100%, they are all increased proportionately. That is happening in some of the other ridings I mentioned above.

      Delete
    4. so according to your model, what does the NDP end up with in St John's East? Just curious.

      Delete
    5. I think the NDP would get 75% in St John's East.

      Delete
  5. 59% in NL and your model only gives them 4 of 7 seats? Gotta question that modelEric!

    JKennethY

    ReplyDelete
    Replies
    1. See my response to Kain.

      Delete
    2. I think the NDP vote is concentrated in St. John's, the Avalon, and Humber-St. Barbe-Baie Verte. These regions only include 4 seats. The other three seats (Bonavista-Gander-Grand Falls-Windsor, Random-Burin-St Georges's, and Labrador) are competitions between the Liberals and Conservatives, the NDP stand no chance in these ridings.

      Delete
  6. Hi Eric,

    I don't know if you've seen it yet, but the provisional boundaries are out for Newfoundland, New Brunswick and BC.

    http://www.redecoupage-federal-redistribution.ca/content.asp?document=home&lang=e

    ReplyDelete
    Replies
    1. I have, but I'll wait for the official boundaries and the transposed results. Things could still change.

      Delete
    2. Just looked at B.C.'s more closely - that is a lot of change! Almost every riding has been carved up, and a lot of the new ones straddle multiple old ridings. Will make for some interesting nomination battles.

      Delete
    3. Those are some interesting provisional boundaries for NFLD.

      Delete
    4. I know there have been some discussion(Huffington) about how the ridings are being "physically" shaped in places like Regina!

      The pie shaped rural/urban ridings tend to favor the CONs by distributing the urban votes over multiple ridings splitting small slices of urban voters versus larger chunks of rural voters!

      EM

      Delete
    5. In other places like Vancouver the creation of vancouver-Granville likely gives the Liberals a seat and the redistricting of New Westminster, Burnaby and North Vancouver ridings is favourable to opposition parties.

      You infer the Tories have usurped the process to gerrymander ridings. That is a serious charge since, you accuse a Judge of the Court of Queen's Bench, A former professor at the University of Saskatchewan and the presidsent of the Saskatchewan Association of rural Municipalities complicit in reorganising ridings to favour the Conservatives to gerrymander and subsequently steal the next general election.

      Do you have evidence to support your slander?

      Delete
    6. I see no such inference in that post.

      Delete
    7. Eric,

      Perhaps my response to Anon June 30 9:42 "EM" was a little harsh. I apologise.

      Delete
    8. It's hardly controversial to point out that the demographics of the pie-slice ridings in Saskatchewan favour the CPC. To note that doesn't mean that the commission creating the boundaries is seeking to have that effect. There are arguable benefits in terms of higher social cohesion to having ridings that combine urban and rural. I don't buy the argument myself, but I've heard it made by reputable sociologists.

      Delete
  7. "The Conservatives have dropped 15 points in Manitoba to 39%, while the NDP is up nine points to 35%. The Liberals, meanwhile, are steady with 17% support. This would likely deliver four seats to the NDP and one to the Liberals in the province, with the remaining nine going to the Tories. That is down two seats for them."

    The "four seats" the NDP shall receive, I assume, would be Churchill, Winnipeg Centre, Elmwood-Transcona, and Winnipeg South-Centre? Or would Winnipeg South-Centre go Liberal red?

    Maybe there's not enough resolution in the sample to pick out ridings, but I assume that most of the Manitoba gains would be in Winnipeg.

    ReplyDelete
    Replies
    1. I'd guess Winnipeg North before South-Centre.

      Delete
    2. Or maybe both?

      Delete
    3. I'm not entirely sure on how Eric's model work, but I assume it's based on some general, "in most cases" works, vote total to seat allocation. The thing that I'm curious about is pretty micro-level stuff, but I'm really wondering if rising NDP counts in Manitoba are due to

      1) Liberals in Winnipeg South Centre and Saint Boniface abandoning ship and accepting the NDP as the party of the centre-left, as they have at the provincial election level.
      2) Just rising NDP votes everywhere rather uniformly (in which case, a lot of the rise would be "wasted" but it might tip the next challenger over the edge of Kevin Lamoureux).

      Kevin Lamoureux is quite an excellent, one of a kind, campaigner for the Liberals - he has a solid constituency in the Maples that contributes substantially to him getting elected. If the Liberals get one seat in this projection, would it be by retaining lamoureux's Winnipeg North riding or by regaining Saint Boniface or Winnipeg South Centre?

      Delete
  8. Where did you get information sub-regional percent in Environics? I can't find them.

    ReplyDelete
    Replies
    1. As mentioned, they were provided to me personally.

      Delete
  9. Never has there been a better argument for getting rid of the outdated FPTP system. For over a month now, the NDP have consistently held nationwide leads over the Tories yet the Cons continue to project winning 10-20 more seats anyway. Absolutely ridiculous.

    ReplyDelete
    Replies
    1. Referenda were held in a number of provinces all rejecting electoral reform. Proportional representation presents its own problems.

      I would point out that PR would likely cause the NDP to lose seats. 35% nationally only equates to 108 seats. Even on a provincial basis the NDP would lose seats 27 alone in Quebec with the poll numbers.

      Delete
    2. I am not really out for a system that will benefit the NDP or any other party, I am out for a system that will reflect the will of the voters. If you have the plurality of votes, you should have a plurality of seats, simple as that.

      Delete
    3. Hurtin Albertan,

      There are lots of ways to reflect the will of voters; at present we do so along geographical lines. In order for votes to reflect seats constitutional amendment would be required as seats are distributed via provinces.

      One reason electoral reform failed in BC was due to geographical concerns. The second referendum on BC-STV oulined reformed ridings. The city of Vancouver was a single constituency with multiple members, neighbourhoods and communities were not represented. The tradeoff for seat proportionality was loss of identity and local representation. This was as big a concern in urban ridings as it was in less sparesely populated areas. People like to know who their representative is-who is accountable. Time and again the issues of access and accountability arose. People were unwilling to give up guaranteed representation for proportionality in the Legislature.

      Delete
  10. Just because Eric's projection model keeps giving more seats to the Conservatives even when their popular vote is projected to be lower than the NDP's doesn't mean FPTP is the culprit. Eric's model is just one person's formula for guessing what the seat distribution is likely to be. Seat projection models are a dime a dozen....and each one has different formulas and assumptions. I think that in the end, the party with the most votes will have the most seats.

    ReplyDelete
    Replies
    1. Yep, just like the Quebec Liberals in 1998, or the BC Liberals in 1996! Or when the PCs won two seats in 1993 with 16% of the vote, compared to 52 for Reform with 19%.

      As has been shown time and time again, the model can turn regional/provincial vote totals into reasonably accurate seat totals.

      The NDP's national lead is irrelevant - how their vote breaks down regionally is extremely important.

      Delete
    2. Eric's formula has the benefit of being based on math and numbers rather than intuition. Once you start putting your finger on the scale to alter how the results come out and make them fit your desired outcome, then you've tossed out any claim to projecting. All you are doing is wishing.

      Also, it simply isn't a straight line correlation between most votes and most seats. Think of elections like the 1979 federal election. Trudeau's Liberals won 40% of the vote, compared to 36% for the PCs, and yet the PCs still won more seats. If that could happen, why would it be surprising that the CPC could win a plurality of seats despite losing the popular vote by 2-3%?

      You especially have to take into consideration how finely tuned the CPC microtargeting machine is. They are focused on getting relatively narrow wins by targeting specific small segments of the population. They are doing it for precisely the reason that a first-past-the-post system rewards it.

      Delete
  11. And then there is handing the election to your opponent !!

    http://www.thestar.com/news/canada/article/1219932--cohn-tim-hudak-s-tory-vision-for-a-low-union-low-wage-ontario

    ReplyDelete
    Replies
    1. Ontarians hate Tim Hudak. I think the next election in Ontario would bring up a Liberal majority government, with the NDP as the official opposition. The PCs would be shut out.

      Delete
    2. Hudak's proposal is a good one, though I don't understand why workers who choose not to join the union should receive union benefits. I'd suggest he should mandate that all shops be open and thus allow non-union workers to work alongside union workers. If the unions offer benefits that are sufficiently attractive, then the workers will join. If the unions do not, then they will lose members.

      It's not unions that are the problem - it's closed shops.

      Delete
  12. Of course if the popular vote in a province is so close as to be a virtual tie...it's a coin toss as to which party gets more seats. But in cases where you have a party getting more seats with fewer votes, it's usually because the party that wins the popular vote has a "wasted vote" problem of winning certain ridings by massive margins and losing narrowly elsewhere. The classic example of this is with the Quebec Liberals winning as much as 90% of the vote in some non-francophone ridings. I don't see anything comparable in Canada apart maybe from the Tories wasting a lot of votes winning seats in Alberta by massive margins.

    ReplyDelete
    Replies
    1. It is death by a thousand cuts.

      In B.C., the NDP needs to have a wider margin over the Tories in order to start winning a lot of seats. Their return on being up a few points is quite low.

      In Alberta, their gains are wasted entirely.

      In Ontario, the NDP's return on gains is also very low, particularly as most of the Tories' close races are CPC/LPC ridings, meaning the NDP has a lot of ground to make up.

      The NDP wastes votes in Quebec just as the Tories do in Alberta, but there are more seats for that vote to be wasted in.

      And in Atlantic Canada, the NDP's vote is very inefficient and depends on the Liberals losing a lot of support.

      The numbers are the numbers, they cannot be wished away. Certainly, the NDP could do better than what projection models give them, but it is unlikely to be so wrong to the extent that 10-15 seats switch exclusively from the Tories to the NDP.

      Think back to the Manitoba election - my seat projection model showed that the NDP could win a very comfortable majority even if they were neck-and-neck with the PCs, and even if they were trailing by a few points. It isn't a "coin toss", there is a science behind these models.

      Delete
  13. We are also going into uncharted territory when you have a party becoming competitive for the first time in a new region...I think what you will see in Ontario next time is the NDP vote going way up in places like Brampton where they will essentially Hoover up most of the old Liberal vote. I don't care what your seat projection says if theNDP takes 59% of the vote in NL they will win all seven seats in the province. At that level even if Jack Harris and Ryan Cleary won 100% of the vote in their ridings, there would still be enough votes left to win the rest of the province.

    I don't have problem with people doing seat projections, but in the end any seat projection is nothing more than a HYPOTHESIS that may or may not be borne out and there are many people doing seat projections all using different formulas and calculations etc...and all getting different results. We can speculate on which party's vote is more efficient but it's all just hypothesizing unless someone polls 500 people in each of the 338 ridings.

    ReplyDelete
    Replies
    1. DL, if my seat projections are a hypothesis, your estimate of what will happen in Brampton and NL, etc. is a guess! At least I have something that is demonstrably effective in turning regional numbers into seats.

      All the seat projections I have seen have mildly different results - not greatly different ones. Most show the same thing, that the NDP is not going to be winning the most seats with a tie or narrow lead in the popular vote unless the regional numbers change.

      Different models would have shown the same kind of thing in Quebec in 1998 and, if they were done correctly, Manitoba in 2011.

      Delete
    2. DL, you seem to be determined to ignore what Eric is saying. The model has, in fact, been able to predict what would happen when a party experienced an unprecedented surge of support in some previously barren area. It did a pretty decent job of calling what happened in Quebec in May 2011.

      We really also can't say where the NDP will go up in Ontario without sub-regional polling. It's entirely possible that the NDP has become highly dominant in the 416 and Northern Ontario, while taking a lead in South-Western Ontario.

      In such a situation, the gain in Northern Ontario would put, at most, three more seats into play (Sault Ste. Marie, Kenora and Nippising-Timiskaming).

      The gain in South-Western Ontario would probably put three to four seats into play (London West, London North-Centre, Essex and just maybe Chattam-Kent-Essex).

      In the city of Toronto, I would expect a surge to put into play perhaps five seats (Scarborough Centre, Scarborough Guildwood, York West, Don Valley East and Toronto Centre).

      So from those three regions, where based on traditional patterns of support it is most likely that the NDP is spiking, there are 12 seats that could reasonably be delivered to the NDP. There are, of course, the occasional other as well. I think it is likely that the NDP will pick up Bramalea-Gore-Malton, and has an outside chance at Brant and Cambridge. But even putting all of that together, that's 15 seats more for the NDP for an Ontario total of 41. An important improvement, yes, but not half of the seats in Ontario.

      Of course before we can really understand what the effects will be, we have to wait to see 1) whether the NDP numbers in Ontario will be sustained and 2) what the new riding boundaries are going to look like. The riding boundary change could have dramatic effects on this math.

      Delete
  14. I'm not at all surprised by the NDP weakness in Saskatchewan. Now that Saskatchewan has a taste of prosperity and the policies prosperity requires, they're not eager to throw those away by voting for socialists again.

    ReplyDelete
    Replies
    1. Then again, maybe SK will get tired of right wing parties when 2015 comes. Maybe then they would vote for a left wing party. Anything can happen in the next few years.

      Delete
  15. Saskatchewan got its prosperity under the NDP - the economy was already booming when Wall won in 2007. btw: we should note that the Environics sample sizes in MB and SK are very small and should be taken with a boulder of salt.

    Actually Eric's model was not particularly good at calling what happened in Quebec - he projected the NDP to win 60-odd seats nationwide - not 103...I understand that the model has since been refined. But in any of these seat projection models - you are only ever as good as your last election. I wonder what the seat projections would have been in Ontario in 1990 when the NDP took 37% and the Liberals 32% and the PCs 24%...I suspect that no model would have given the ONDP anywhere close top the 74 seats they actually ended up getting.

    One of the biggest problems is that Ontario is so damn big that the province wide numbers tell us little about what is happening in each region.

    So again, its a fun parlour game to speculate on how many seats each party will get looking at the nation-wide numbers - but when all is said and done in 2015 there will be 338 separate elections held in each riding.

    ReplyDelete
    Replies
    1. Come on DL, you're being deliberately contrarian. The seat projection model gave the NDP 60 seats in Quebec with the actual provincial vote plugged in. This is about being able to translate regional vote totals into seats, not about being able to predict those regional vote totals.

      Delete
    2. It's all here:

      http://www.threehundredeight.com/2011/05/projection-vs-results.html

      (And I didn't give the NDP 60-odd seats in 2011, I gave them 78. But that error was due to a misreading of the polls, it had absolutely nothing to do with the seat projection model.)

      Delete
    3. Which government implemented the business-friendly policies is irrelevant. Saskatchewan people now see the effects business-friendly policies have, and the current NDP opposition isn't promising anything of the sort.

      Delete
    4. Which party implemented the business-friendly policies is irrelevant. Now that Saskatchewan people have seen the effects business-friendly policies have, they appear to favour parties that offer business-friendly policies, and that does not describe the current NDP opposition.

      Dwain Lingenfelter called for all wages (even private sector wages) in Saskatchewan to be determined by a government office to ensure fairness. Wage controls are not business-friendly.

      Delete
  16. If you can demonstrate that your seat projection formula could have predicted the number of PC, NDP, Bloc and/or Reform seats in 1993, then you're talking about something comparable.

    As it stands, the party system is undergoing a massive shift, plus incumbency won't be worth nearly as much in 2015 on the new boundaries.

    59% of the vote producing only 4 out of 7 seats in a three-plus party system in Newfoundland and Labrador is absurd.

    ReplyDelete
    Replies
    1. The model was able to translate the 12% to 43% for the NDP in Quebec into 60 seats (real result: 59) and Wildrose's 7% to 34% in Alberta into 18 seats (real result: 17). It has no problem with massive shifts that are very different from prior elections.

      It also has no problems with new boundaries: 56 out of 57 ridings were correctly called in Manitoba.

      As I mentioned, with these numbers in Newfoundland and Labrador it is a close race in the three other seats - and close enough that one or two of them could very easily go to the NDP, against expectations.

      Delete
  17. Hey DL:

    Eric can't compensate for bad polling.

    Why at this point in time are you so upset about the polls and Eric's interpretation of them? Even if the model showed the NDP in the lead in seats there is a long time before an election. Personally I think Eric does a great dealing with the polls. He's more accurate than anyone else I've read.

    At this point in time the polls are telling me that there is some concern with the way Harper is operating with his majority. There is also some discomfort over the accusations about the robo calls. I share both of these concerns. As someone who supports Harper I have to say I'm disappointed in him so far.

    Finally DL simply because Eric's results aren't what you want, doesn't give you the right to climb all over him. Lay off my friend!

    ReplyDelete
    Replies
    1. I'm not climbing all over anyone. I respect what Eric is doing here and it is interesting. I'm simply pointing out that seat projections are simply just that - PROJECTIONS. They are based on hypotheses and they are not infallible. It has nothing to do with what results I "want" to see in the next election. I wish the seat projections in Alberta and Newfoundland in the recent provincial elections had been right and the provincial Liberals had been wiped off the map - but in both cases hey got a lot more seats than any seat projection would have predicted.

      Delete
  18. It really would make more sense to put the Bloc between the NDP and the Conservatives on the pie graph, rather than between the NDP and the Liberals. It would make it easier to see when the NDP/Liberals/Green could come up with a majority, since obviously no one would ever rely on the Bloc for a stable majority. It would also give a better view of common voting patterns--NDP and Liberals may vote together, NDP and Bloc may vote together, NDP and Liberals and Bloc may all vote together, and Bloc and Conservatives may vote together (assuming the issue is right). The current configuration doesn't allow for you to see all that as easily. I understand it's meant to approximate a left-right spectrum, but neither the Bloc nor the Greens work on the left-right spectrum, so why force it?

    ReplyDelete
  19. Bev Oda resigns as Conservative MP for Durham

    Toronto Star

    ReplyDelete
    Replies
    1. Finally! Now it's Gerry Ritz, Peter Mackay, Tony Clement, John Baird, Jim Flaherty, Peter Van Loan, Dean Del Mastro, Peter Kent, Christian Paradis, and Stephen Harper's turn to step down next. :)

      Delete

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.