Rubber Hits the Road for Parties … and Seat Projectors
This may surprise some people, but very few of the current amateur seat projection websites have even a single federal general election track record under their belts. And none of them has had to predict an election where so many assumptions have been upended, and so many tectonic shifts have been telegraphed in leading indicators whose full effect has yet to be seen in the horserace numbers.
Here was the track record of the final 2008 seat predictions from various seat predictors in the field during the last election (thanks to Paulitics for archiving them, since many of the authors have subsequently taken theirs down):
| Cons | Lib | BQ | NDP | Oth | Grn | VAR | |
|---|---|---|---|---|---|---|---|
| 2008 GE actual | 143 | 77 | 49 | 37 | 2 | - | – |
| Ekos | 136 | 84 | 51 | 35 | 2 | 0 | 18 |
| LISPOP | 135 | 87 | 51 | 33 | 2 | 0 | 24 |
| Trendlines.ca | 132 | 84 | 52 | 39 | 1 | 0 | 24 |
| H & K 5-poll avg | 129 | 82 | 54 | 42 | 1 | 0 | 30 |
| Paulitics method I | 129 | 86 | 55 | 38 | 1 | 0 | 31* |
| Paulitics method II | 127 | 83 | 55 | 42 | 1 | 0 | 34 |
| democraticSpace | 126 | 92 | 52 | 36 | 2 | 0 | 36 |
| Boragina** | 124 | 88 | 51 | 42 | 2 | 1 | 38 |
| Election Prediction (EPP) | 125 | 94 | 51 | 36 | 2 | 0 | 38 |
| r^2 (Party vs VAR) | 0.959 | 0.328 | 0.026 | 0.176 | 0.006 | 0.167 | |
| * Note, original projection added up to 309, possibly for methodological reasons. ** Now known as Riding by Riding. Uses the downloadable "Elect-o-matic". |
|||||||
96% of the variance in the last election's seat projections could be explained by differences in prediction of the number of seats won by the Conservatives.
This was caused mainly by differences in predictions for the Liberals (explained 33% of the difference), followed by predictions for the NDP (explained 18% of the difference), and by predictions for the Green Party (explained 17% of the difference).
Significantly, not one seat projection methodology over-predicted the Conservatives, and not one under-predicted the Liberals or Bloc Québécois, although there were predictions on either side of the NDP's final total. Only one predicted that Green Party Leader Elizabeth May would win Central Nova, NS (she didn't), while 4 of the 9 methodologies missed predicting both Independent candidate Bill Casey's win in Cumberland-Colchester-Musquodoboit Valley, NS and André Arthur's win in Portneuf-Jacques Cartier, QC.
The inherent bias in seat prediction methodologies to favour previous election results means they tend to overly favour parties set to lose seats, such as the Liberals and Bloc Québécois in the last election. They also tend to miss the likelihood of parties on the rise to gain seats, such as was the case with the Conservatives in the last election. Only the NDP, whose vote intention numbers showed little gain by the end of the 2008 campaign, saw seat count predictions on both sides of its eventual total.
Another problem for the seat projection methodologies is that they are backward-looking. They're using days-old polling data at a time of incredible movement in the polls, and laying that on top of results from the last election when incumbency was a factor for some political parties' votes that is no longer at play.
Moreover, they can't account for turnout, in the sense that parties with momentum, or who have strong on-the-ground organization, will experience higher turnout of their own vote, than will parties who are organizationally weak and/or whose supporters are feeling demoralized.
The above problems were underlined in ThreeHundredEight.com's projection/prediction for the 2010 New Brunswick general election, one that saw a one-term narrow Liberal majority turfed in favour of a massive Progressive Conservative majority government. His then-methodology over-predicted the Liberals by 10 seats (23 versus 13) out of 55.
Since then, ThreeHundredEight's sensational projections have predicted doom and gloom for the NDP on the front page of the Hill Times (no, he's not a "pollster" as they wrote) and the Globe and Mail. As recently as late January he claimed they would lose 13 seats, upped to 16 seats by early February, which emboldened some Liberals to predict they could gain 100 seats during an election campaign.
It took some peer review to examine his original methodology and determine that he had in fact placed a cap on the number of seats a party could be projected to win in any region (equivalent to the maximum it had even won plus those it came within 10% of winning), but put in place no comparable floor. Clearly the wrong assumption for the current election!
Apparently the methodology has since been changed, but not before it set the entire frame of coverage by the Parliamentary Press Gallery for the period leading up to and just following the federal budget vote ("NDP weakness sets up two-way race between Harper and Ignatieff").
Indeed one could say that this one blog – without a single federal general election's track record to its credit – was responsible for the mass failure of the Ottawa punditocracy to foresee either the NDP's willingness to or interest in voting down the budget at the end of March, and for all we know the willingness of the Liberals to provoke an election dating from around that time.
So, what does all that mean for the current election?
It means that:
- the Liberals and Bloc are still likely being overly favoured by all these seat prediction methodologies,
- the projection methodologies are going to wind up missing NDP gains, particularly if the party continues to climb in the polls, and that
- projected seat counts for the Conservatives will likely fall on either side of their final tally.
It also means that the "strategic voting" websites, who are basing their recommendations on seat projection/seat prediction methodologies like 308's are likely making a number of erroneous recommendations — another reason to be rid of those undemocratic and irresponsible projects once and for all — and that people voting in the Advance Polls shouldn't put a lick of confidence in them, as a result.
Let's look at the latest seat projections from each of the contestants in the current election:
| Cons | Lib | BQ | NDP | Oth | Grn | Date | |
|---|---|---|---|---|---|---|---|
| 2011 GE actual | ?? | ?? | ?? | ?? | ?? | ?? | May 2 |
| 308 | 150 | 76 | 45 | 36 | 1 | 0 | Apr 23 |
| Calgary Grit | 150 | 74 | 48 | 35 | 1? | 0 | Apr 18 |
| Cdn Elec Watch | 152 | 71 | 41 | 43 | 1 | 0 | Apr 24 |
| democraticSpace | 157 | 69 | 42 | 39 | 1 | 0 | Apr 23 |
| Ekos | 134 | 82 | 32 | 60 | 0 | 0 | Apr 21 |
| EPP* (61 no call yet) | 119 | 61 | 38 | 29 | 0 | 0 | Apr 24 |
| FairVote/Postmedia* | 201 | 53 | 4 | 48 | 0 | 0 | Apr 21 |
| H & K/latest Nanos | 147 | 77 | 38 | 44 | 2 | 0 | Apr 24 |
| LISPOP | 149 | 68 | 39 | 52 | 0 | 0 | Apr 22 |
| Riding by Riding | 147 | 65 | 42 | 53 | 1 | 0 | Apr 24 |
| TheMace.ca | 131 | 73 | 22 | 81 | 1 | 0 | Apr 21 |
| Too Close to Call | 145 | 74 | 42 | 47 | 0 | 0 | Apr 22 |
| Trendlines.ca | 154 | 78 | 46 | 30 | 0 | 0 | Apr 15 |
| High | 157 | 82 | 48 | 81 | 2 | 0 | |
| Low | 131 | 65 | 22 | 30 | 0 | 0 | |
| * considered outliers, for various reasons, and thus excluded from Highs and Lows | |||||||
Two of the above predictions have been excluded from my high-low rankings: (i) ElectionPrediction.org's which does not have a prediction for all 308 seats yet (normal practice of their site), and (ii) the FairVote.ca / Postmedia projection, which deserves a raspberry or two all its own.
It seems that someone at Postmedia entered the results of the last Ipsos-Reid poll into the FairVote.ca "seat calculator". The Bloc had 6%, which the stupid calculator projected out at 4 seats. Clearly, the Fair Vote calculator did not take into account that the Bloc only runs candidates in 75/308 ridings !!
The same Ipsos-Reid polls results of 43-21-24-6-2 (already an outlier compared with other public domain polls, but consistent with their own firm's trends) produces a projection of 181-46-48-0-31-2 when run through the Hill and Knowlton 2011 Election Predictor, with no other adjustments.
So, one raspberry to Fair Vote Canada for putting up a cutesy gimmick to promote electoral reform that has wound up proving they don't know what they're talking about; and another raspberry to Postmedia for not doing a common-sense reality check on those numbers. The results are completely farcical as anyone can see. No, it's not an excuse that journalists can't do math, either.
The Postmedia "reality checks" on how party platforms add up or not, will have to be read with new scepticism.
UPDATE: Of course we should have known that Glen McGregor would have been on the case already.
As for Trendlines.ca, Freddy Hutter tells me their latest projection has the NDP at 74 seats, but I'd have to subscribe to get the data any more than 9 days behind. Oh well … guess we'll have to wait.
UPDATE: From a more experienced point of view, Elly Alboim's blogpost on the recent surge is well worth reading, if you haven't seen it alreaday.
Tags: Seat Projections
No offence pg, but some of these seat projections are just pure unadulterated crap. Here are three reasons for this:
1) We don’t have nearly enough polls, or riding level data to accurate predict the results (Contrast this with FiveThirtyEight in the US which has tons of polls with lots of detail)
2) Most Canadian seat projections are very opaque about their methodologies and those that do suggest what they use for methodologies are not very rigorous (again compare this with Nate Silver’s weighted Monte Carlo simulations and forecasting techniques)
3) Elections here can be decided by 20 votes, whereas in the US (where this is done better), most candidates win by much larger amounts. The law of large numbers thus further favours the US for averaging out measurement and forecasting error.
Finally, I’m disappointed by the uncritical and unprofessional way that these projections are bandied about by the public and members of the media. Canadian “forecasters” should be more forthcoming about their methodologies and people should think twice before accepting this projections.
Thank you again for another stimulating article.
It should be pointed out that the seat predictors are only half to blame for their ‘08 results.
Public pollsters take the other half of the blame.
Junk in, junk out.
The reason for this:
“Significantly, not one seat projection methodology over-predicted the Conservatives”
is because on average pollsters UNDER-estimated Tory support.
Take EKOS which nailed NDP and LPC numbers but had a -3 CPC and +3 Green variance from reality.
To this day Frank Graves is on the high end for the Greens and the low end for the Tories when compared to his peers.
An interesting exercise would be for all seat projectors to run previous elections through their model using the correct values for party support.
At least ‘04, ‘06, and ‘08 where we’ve had the current party configuration.
That might be more revealing.
The PPG needs to take some serious lumps here.
Having one pollster dominate the conversation is bad because quirks in their methedology can give a misleading impression of events on the ground.
Having one forecaster dominate the conversation is equally bad.
Especially when that forecaster teams up with one of Canada’s largest media companies who’s reporters start shaping their stories around the polls and projections while ignoring the reality on the ground.
Alice thank you for pulling a Joan Bryden.
Hopefully this topic will get picked up or at the very least prompt some introspection in the media about the “filter” they’re using to interpret political events.
Caveat Emptor.
Remember the 2010 British election whereby the 3rd place Liberal Democrats experienced Cleggmania after the debates?
The LD’s eventually (and incredibly) vaulted into first place with 34% in a YouGov poll.
Alas, on election day the LD’s finished 0.9% above its 2005 polling mark at 23% but still lost 5 seats in the process – going from 62 to 57 seats.
Just something to keep in the back of the mind in this eerily similar NDP trend-line.
Todd, I realize a number of people have been making those parallels between the apparently “orange wave” here, and what was seen with the Lib Dems in the UK election last year. Before I would draw any conclusions on the lessons that could be drawn from the other case, I’d want to know more about the Lib Dems’ campaign organization, and whether they were as well (or better, or worse) prepared to try and catch a wave than what the NDP believes it is right now.
Cheena, no-one at this blog is trying to take control of your mind and stop you from thinking for yourself and voting according to whatever methodology you like. To the contrary.
My point is simply that these projections are not gospel, not that people shouldn’t vote strategically if they want to. In most cases, people in an area have a perfectly good idea of whose campaign is stronger, and don’t need someone from Toronto to tell them how to vote to achieve whatever objective they want to.
However, more and more people are coming around to the view that “strategic voting” isn’t so strategic, and is also “corrosive to democracy”, for example David Olive at the Toronto Star:
http://thestar.blogs.com/davidolive/2011/04/a-vote-for-jack-is-a-vote-for-jack-period.html
I assure you, the only thing that would worry me is if people weren’t voting.
Thank you both for taking the time to leave a comment.
Fair Vote Canada has no relationship with Postmedia, and we have made no predictions on the outcome of this election.
Fair Vote Canada’s Vote Calculator, which we describe on the site as a “game”, was intended as an educational tool to show how our current voting system wildly distorts election results. It was not expected that anyone would take it seriously as a tool for predicting the results of the election, and we have added a disclaimer to that effect.
That said, the output from the calculator is not totally unreasonable if you accept the validity of the poll that was used as input by Postmedia. That poll showed the Conservatives at 43% and the NDP and Liberals split evenly at 25% each. (Other polls released on the same day show the Conservatives at 36-39%.) Those vote percentages would certainly result in a Conservative majority, if not a landslide.
The projection of four seats for the Bloc is certainly shakier. As noted, our little calculator does not take account of regional polls. With 27% of the votes in the province, the Bloc could still win 30 seats or more. We know that, and would never issue any statement like the one issued by Postmedia and attributed to us. Note, however, that 6% nationally for the Green Party gets you zero seats, not four.
Serious seat projection is done on a riding by riding basis. We are not in that business, and we are distressed that our position was misrepresented by the article in the Ottawa Citizen. We would be grateful if Pundit’s Guide would not perpetuate the myth that Fair Vote Canada is predicting the outcome of the election.
Wayne Smith
Executive Director
Fair Vote Canada
On the other hand, if the Bloc end up with four seats, remember, you heard it here first!
Alice;
I don’t understand the reason(s) you say “projected seat counts for the Conservatives will likely fall on either side of their final tally.” I missed what has changed that would make the projections more accurate for the Conservatives this election? Are you saying they should be more accurate?
Sincerely,
Biaced
Wayne, I think I understand perfectly well what transpired in this case. No-one thinks you have any affiliation with Postmedia.
What’s evident is the naïveté of Fair Vote in putting such a faulty “seat calculator” onto its site, and not thinking that people might use it to … I don’t know … calculate seats. Calling it a game doesn’t get you off the hook, sorry mate.
The Bloc numbers are wrong and wrongly calculated; and you know it, and you’re making it worse by dancing around this.
Moreover you folks were warned not to put a seat projector onto your site, as their methodology in general is so faulty. I don’t know how many more communications problems your cause can overcome. I support MMPR with open lists, but every time I hear about “wasted votes” or now encounter seat calculators that are just wrong, it makes me think that the organization advocating for it will never win this fight for public opinion.
You can’t put out something that wrong, and then hide behind “the Ottawa Citizen misrepresented it”. They used it for its stated purpose: to calculate the number of seats that a given distribution of vote shares would produce. Now, they should have realized just how wrong-headed the answer is, I also argue, but that doesn’t let you folks off the hook in the first instance.
Time to go back to the drawing board on your communications planning, I think.
Because, Biaced, according to my interpretation of what happened in 2008, a party whose vote-share has been stable throughout the campaign, and from one campaign to the next, will be more likely to see a normal distribution of seat predictions.
Whereas a party that’s in decline will be over-predicted, while a party that’s rising (as the Conservatives did last time, and the NDP is so far this time) will be under-predicted. According to my current hypothesis. You may see things differently.
Thank you for the mention. I did make projections in 2006, they are recorded on the old DS website except for Quebec, where due to a personal emergency I was unable to complete all ridings in that province.
As for my seat projections, they are just that, projections. How I expect things to end up. I am not magic, so I do not know for sure, and I can assure you, I will be watching TV on election night, and may have as many as 2 TV’s and half a dozen net feeds going at one time.
As for methodology, I agree there are some lacking websites. 308 discusses their weighting very well but not how they translate that into seats. I’ve always been open as to how I translate my province-wide results into seat results, but my weighting is somewhat lacking, with each new poll pulling the results in each province towards it by a value of 1/5th.
I am not out to attempt to become a God. Back in 2008, only DS and EPP (along with myself) offered riding by riding projections, and I held my own against both of them. I strongly feel that I know what I am talking about when I explain how to translate a province-wide provincial polling number into results for each individual riding, and I wish to prove such. Having access to this information is useful for political parties, candidates, the media, and even voters. That is why I’ve been as open as I can be.
In my humble opinion, but offered with somewhat less opinion:
No matter how complwtely expressed and/or well thought out the various predicting methodologies… its garbage in and garbage out.
And that is the way it will be until you figure out at the very least hoe to FULLY captture the effect of changes in progress, and hopefully the effects of major changes in single riding dynamics.
that was supposed to be ‘offered with less hunility’.
Ah well.
Thank you, Alice.
So, if I understand, the Conservatives rose in the polls during the 2008 writ period and the projections were unable to capture the increase? This election, because polls have generally shown the Conservative support as flat, the seat projections should be more accurate? Does this require an assumption that the bias that appeared across polling firms in the 2008 election consistently showing the Conservatives lower in their polls than the vote percentage tallied on voting day no longer exists?
Sincerely,
Biaced
The Conservatives jumped at the last moment in the polls, but even that did not indicate the size of the win.
No, there’s an additional complication … the Liberal vote was less motivated in 2008, while the Conservatives had a better organizational ability to identify and get out their support.
Presumably when the Conservatives get to that point in the electoral cycle, they will be in the same situation as the Liberals were this past time.
Who knows who is going to wind up in what position by the end of this election!
I did projections in 2008 and was basically on par with LISOP. We, like everyone else, had the Conservatives low because they over-performed against the polls – I don’t think it was due to any inherent problems with the model, since I re-ran my projections against the actual data after the election and the Tory seat total was quite close to the mark.
But I agree with most of what you’re saying. These things should not be take as predictions…they’re based on current poll results. Also, the current NDP surge in Quebec complicates matters imensely – these models all work better when faced with minor, rather than major shifts.
The accuracy of seat projections and opinion polls has been validated in jurisdictions where advance poll results are released. The failure of projections to catch Election Day sentiment is four-fold: a) no polls to digest in the final four days of a campaign, b) dynamics of the ground game, c) determining which supporters will stay home, & d)predicting how the Undecided will peel off.
Polling has been within 1% in recent elections and riding conversions are usually accurate to 10%.
Trendlines has charted key elections across Canada & around the globe since 2004. In the headline monthly chart of Canada’s Federal Elections, our own forecast is blended with the efforts from up to 14 recognized Cndn & international practitioners. During campaigns, our familiar multi-model chart transitions from monthly to daily reporting.
Our methodology differs from the “hobbbyists” in that we often project momentum to “Election Day”. This unique number is made available to policymaker & stakeholder subscribers along with our forecasts for future G-20 Recessions, crude oil Prices and co2 Emissions. As PG mentioned, all charts are made available to the public with a reasonable “delay”.
The real poll will be on the night of May 2nd, especially in an electoral system where you have 308 small elections and where local factors are sometimes very important.
However, even if exit polls are somewhat quite accurate in a few cases like in presidential elections there is a few exceptions. For example, in the UK, the 1992 election exit poll was totally flawed and this even that in the UK it is quite possible to do a sample based on battleground or target seats because they are much more of these seats in number than in Canada.
Even with this, ironically I believe that counting election signs in some areas is better than the ‘’science” of some riding by riding predictions sites. In other words, the ”uniform swing” factor based on regional number does not work in the Canadian electoral system!
And again, like said earlier, the Lib Dem factor is quite important especially in a multiparty system with the FPTP system. In 1997, the PC Party had a much higher score in the polls
Also, in the 1988 election, the Liberals were going stellar in the polls in mid-campaign, but the PC won a majority at election night.
Ekos’ anti Tory bias shows even more in seat projection.
CPC 131 LPC 62 NDP 100 Bloc 14 Other 1
Before I would draw any conclusions on the lessons that could be drawn from the other case, I’d want to know more about the Lib Dems’ campaign organization, and whether they were as well (or better, or worse) prepared to try and catch a wave than what the NDP believes it is right now.
The short answer is that their national campaign organisation was terrible and that they were totally unprepared to dealing with their surge in the polls.
But a short answer isn’t really enough, so here are some random points in lieu of a proper answer:
1. The LibDems have not had a serious national organisation since the old Liberal Party disintegrated in 1931. During the post-war period they survived because of personal votes and the occasional star candidate and this became habit-forming to the extent that national or regional organisation became associated in the minds of Liberal activists – who from the 1960s began to define themselves (laughably) as anti-establishment/anti-system – with the two main parties and thus as something that was somehow anathema to liberalism. The long-term consequence of this has been a party that has been excellent at exploiting localised discontent, but which has consistently failed to exploit poll surges during election campaigns. In 1974 and 1983 (when, unlike 2010, there was an *actual* surge in Centre support at the ballot box) they ran up impressive percentages in totally unwinnable constituencies and failed to win most of their target seats. That pattern was observable, though not so striking, in 2010.
2. It is likely that the apparent surge in LibDem support in 2010 was the fault of incompetence on the part of the polling industry as there is strong evidence that there was no surge on the scale all polls predicted. Postal votes (obviously) had to be returned a few weeks before the election and postal votes for local elections a week before those for the General Election. In Newcastle North (a LibDem target) around 40% of votes were mailed in; the highest percentage in the country. Not only was it comfortable Labour hold, but Labour made large gains in the local elections in the parts of Newcastle covered by the constituency.
3. The LibDems had not expected to make serious gains in the election and was not prepared to run an aggressive campaign. Their targeting was abysmal; resources were switched from vulnerable seats and target seats with small majorities to constituencies that they would not have expected to win absent a large increase in support. A particularly extreme example was Rochdale, where the LibDem incumbent managed to lose despite it being the town where Brown forgot his microphone was still on. In South Wales the LibDems would have gained the Labour marginal of Swansea West had they run a normal campaign, but instead resources were wasted in places like Merthyr. There are countless other examples, as I’m sure you can imagine.
4. The situation was exacerbated by apparent tendency of LibDem activists in previously unwinnable constituencies to attempt to win on their own patch rather than transfer their efforts to a target constituency nearby as they have traditionally done (at least that’s what is reported in various reliable publications; I don’t actually know any LibDem activists).
You get the idea. I’ve no idea how relevant any of it is to the NDP in this election, but those that know can compare.
Though a further point about comparing elections in Britain and Canada has to be made; Canada has one of the most volatile electorates in the industrialised world. Britain is just the opposite; while details endlessly alter (Liverpool was once a Tory city, Norfolk a Labour county), the general *underlying* pattern of partisan support has been frozen since about 1935 or so. Which means that is much, *much* easier for a Canadian political party to translate a surge in support in opinion polls to a surge in support at the actual polls.
Can anyone tell us why Ekos is consistently tracking off against the rest of the polling companies? 100 seats for the NDP and only 131 for the Cons? 3000+ respondents. It is very suspect. This is SUCH an outlier.
George Pringle one of the interesting aspects of election day is the amount of crow that will need to be eaten, be it by Frank Graves, Eric Grenier, or any other pollster/projctor who’s off the mark.
There’s an amazing divergence in numbers. Outside the MOE differences.
Somebody is wrong, let’s just wait and see who it is!
None of us want to be wrong of course :P but it happens.
It certainly is evident that Quebec will play an important role in determining the slow demise of the BLOC. This is good news of course! However, it is somewhat unclear if the surge of the NDP will translate into actual NDP seats. It may fragment the vote in such a way as to have impact across the Liberal vote in a favourable way. The BLOC is bound to loose upwards of 12 seats … NDP may only get an additional 3 .. leaving the lberals with gains also. Good news that the Bloc will likely be in the low 30’s at best !!!!
It is not clear that Fair Vote Canada’s Vote Calculator is “faulty”. You can put whatever numbers you want into it, and if the numbers are sufficiently stupid, it will generate stupid output.
The Ipso-Reid poll numbers that were used in the Postmedia projections are not supported by other polls released the same day, and are looking shakier as time goes by.
In order to accept the seat projection, you have to accept both the accuracy of the calculator and the accuracy of the poll. We are highly skeptical about the poll.
We won’t know whether the Vote Calculator is accurate unless the election results happen to match the poll projections. Of course, once we have the actual election results, we can put them into the calculator and see what happens.
We don’t have a problem with people using our calculator to calculate seats. That’s what it’s for. What we have a problem with is people putting in numbers of their own choosing, then issuing statements that say “Fair Vote Canada predicts . . .”
To reiterate, Fair Vote Canada has issued no seat projections.
Wayne Smith
Executive Director
Fair Vote Canada
ThreeHundredandEight has changed methodology today… while this is clearly “trying to get it right” (a good thing), results over time there are now no longer comparable (a bad thing that surely won’t get mentioned by most media outfits who quote the site … about says it all.
The entire strategic voting biznak has been based on his predictions.
308 is consistently moving more towards a methodology like mine :P
Perhaps Jack is drinking the EKOS polls/projections’ bathwater? By design or purpose EKOS undercalls Tory vote and overcalls Lib vote coincidentally is used to affect the voters. So how is the manipulation of Jack occuring through this poll/sp?
Well, the questions yesterday started the people thinking “Do you want to elect the NDP as the government” and “Do you want another Bob Rae”.
So if you are a rabid Harper hater who is switching from Lib to ND, EKOs is providing “evidence” that as usual is only in the interest of the Libs.
How convenient.
We are having 308 mini elections with a rallies used to energize the locals when leaders and their entourage.
I am going to stick with my Conservative forecast of 159 for the CPC I made on March 23, 2011.
I do think the CPC and NDP have best ground game. I don’t believe they can tap into 40+ new seats. A conversion of dozen for CPC is realistic based on their past performance. I don’t think it is realitic for NDP poll-mania to suggest 60 seats for NDP.
I would have NO problem if voters gave Federalist parties EVERY Bloc held seat though on May 2, 2011.
http://canadiansense.blogspot.com/2011/03/conservative-majority-end-of-political.html
308 and EKOS are taking A LOT of heat on the internet these days, although we should probably wait until after the results are actually in before judging them.
I’m glad Alice is keeping score.
Quite frankly the CBC-Graves and G&M-Grenier relationship is going to be called into question if they are wrong by too large a margin.
QMI is probably licking their chops to go after Frank Graves because he’s every Conservatives favourite boogey man.
More fodder for media criticism and the never ending fight to cut off funding to the state broadcaster !
308 has weighted his new polls at 7% (IE the value at which old ones die)
I use 14%, and used to use 20%. 14% is much better. Originally, from what I could tell, he used 2%, and until today, he used 4%.
My point is 308 is getting better, and while I do take shots at it from time to time, it is a website I visit multiple times a day.
You may think that I am crazy, but don’t be surprised if the NDP will win less seats than in ‘08.
Why? Because the gains in Québec will not compensate for the *riding* losses in Northern Ontario, Alberta and BC.
I know Québec could be awfully volatile at times, but I believe that they will win 5-6 seats at most in Quebec. With this, maybe the CPC could win 3 to 4 more seats in Quebec due to the Bloc going down especially in Eastern Québec, Centre-du-Québec and Quebec City.
However, on the field, Gatineau will go NDP, with maybe Jeanne-Leber and a few other ridings in the 514 and 450. They will probably have an awful night at the end, but the liberals have a chance at winning Ahuntsic and Saint-Lambert as three-way races. Even with this, ironically, Duceppe’s seat could even be more close than excepted just like Rosemont-Petite Patrie.
BTW, could Ignatieff lose his own riding? It’s a very long shot, but it’s not out of the question.
I did have a prediction for the 2008 election, which I made Oct 3
Cons 164, Libs 54, Bloc 52, NDP 38, Ind 2. I admit I over estimated the fall of the Liberals.
http://bciconcoclast.blogspot.com/2008/10/election-preditions-on-october.html
I privately predicted in 2006, I was not yet blogging, I came to 123 Cons, 110 Liberals, 45 Bloc, NDP 28, Ind 1 – I had a lot of analysis done for the UBC ESM I was playing, I made out like a bandit in 2006
I should have added, in April 1999 I predicted the April 2001 BC election to within 2 seats. I had it 71 Liberals and 4 NDP – I thought they would win North Coast and Powell River Sunshine Coast, which they did not.
I did that work for former NDP cabinet minister Bill Barlee
Do you have a prediction this time, Bernard?
I also noticed your comment on Bill Tieleman’s blog, where you said some surprising seats would be in play in BC with the NDP at these levels? Care to share your thoughts on which ones those would be? I have some ideas of my own.
You will get the best projections from a prediction market where you have competent participants. Inevitably I was a little disappointed when the UBC ESM was not open for this particular election. It was a reliable source of money.
I notice today at least some seat projectors have made the adjustments required to reflect the new reality such as electionalmanac.com which are showing the NDP, based on regional numbers, poised to elect 103 seats. And that is even before the Angus Reid poll which casme out last nite showing the NDP at 30% and still climbing in the polls.
Thank you very much Alice for your analysis of what is actually going on.
I fall for the punditsguide a little bit more everytime I read you.
If seat predictions were a science we would not need elections.
I will note that I appreciated electrionprediction.org (when people still used it) as it allowed people on the ground to comment.
In Quebec, it’s all about whether the NDP voters actually get to the polling station. Given the euphoria, ….
tOM
Hi Alice: If you’re not already in line to win the Michener Award for this website, you should be. Thank you for helping average voters like myself make sense of all this.
To see the ‘value’ [sic] of seat projections, remember the 1990 provincial election in Ontario. There were repeated claims the Progressive Conservatives would win only four seats, based on a uniform swing projection from the 1987 totals. Of course, they didn’t win four – they won 20, despite a decline in the popular vote.
There was also that classic moment in one network’s 1990 election night coverage when they flicked up a seat ‘projection’ [sic] that claimed the NDP could not win more than 62 seats…even as the party was in the mid-70s on the leader board (a point helpfully underscored by one of the pundit panelists). The NDP finished with 74, a total that included such ‘impossible’ [sic] pick-ups as Lambton, Frontenac-Addington and Halton North.
And how about that ‘projection’ [sic] that the Ontario Liberals would win 106 seats in the 1995 provincial election?
And does anyone else remember the runaway success of projection-fueled strategic voting in Ontario in 1999? Yeah, that was golden.
Yesterday’s Lambton is today’s Drummond is tomorrow’s Yellowhead. Round and round we go.
Seat projections are like horoscopes: you see only what you want to see and leave out the rest. Unlike the video-game virtual-reality of algorhithms, you don’t get to ‘leave out’ the rest in a real election. As a real voter making a real decision in a real election about real issues that affect my life, I am about eight miles past done with the static noise of ‘projections’ [sic].
Bearpassage, do u recall the caveat phrase “correct 19 times in 20″ at the bottom of all those polls & projections?
Most of the air time has been given to “hobbyists” in this writ, but the accuracy rate is over 90% among the tier-1 practitioners that have weathered several campaigns.
Sigh I guess this is the accepted wisdom now:
“No one saw the NDP surge coming, not even the NDP.” – J.S. in his new column
Actually polling has been around for at least a year showing Layton as the most popular leader in Quebec and having way, way better personal numbers than Ignatieff.
My talk of ‘convergence’ took place well before the election and i’m a committed CPC partisan with no dog in this hunt.
Message to PPG and political columnists:
Just because you failed to see the possibility don’t assume everyone else did as a way to excuse your faulty prognostication.
Hello Alice;
I watched CTV Power Play online where you were explaining how seat projections can under-estimate the party on the move. For example the NDP seat projections could be very underestimated. While the Bloc, Liberals and Conservatives may be greatly overestimated given that they are falling in poll numbers.
My own personal hunch is that there is still a small bias in polling, maybe 3 percent, against the Conservatives. The proof will be in the final poll results vs. the ballot results.
Ah, but they weren’t regular readers of Pundits’ Guide like you, Shadow.
It’s not a polling “bias” per se, I don’t think, Biaced. It’s a turnout difference. Turnout has varied by demographics and been helped by ground game. Two advantages the Conservatives had in the last election.
At midinite, Compas released what I deem to be a rogue poll. CPC 46% !! Because of the ElectionsCanada blackout, no other agency will be able to validate its results.
Stay tuned @ http://trendlines.ca for details & riding projection.
breaking news: if a Compas poll released at midnite Friday with the Conservatives @ 46%, NDP 26% & Liberals 17% is not a hoax, it converts to:
193 MPs – CPC, 52 NDP, 44 BQ & 19 Liberals!
These results cannot be validated by other agencies. ElectionsCanada has an opinion poll blackout for the next 72 hours. I am skeptical of these results and consider this a rogue poll. Compass surveyed 750 on Wednesday & Thursday. This does not compute with similar surveys by Nanos & Angus Reid on the same evenings…
Fortunately I did not realize the blackout period has been reduced to 24hrs. Four new polls will be available by midnite tonite to dispel this rogue poll! Stay tuned to our site for updated projections thru to Monday morning.