Tuesday, June 17, 2014

Guest post: Frank Graves of EKOS Research discusses his likely voter model

The final poll of EKOS Research, whether it be the tally of all eligible voters or those of likely voters, told much of the story: the Ontario Liberals were on the cusp of a majority government. But the poll suggested the NDP would take a significant step backwards from its performance in the 2011 election, which turned out not to be the case. Why?

In this guest post, Frank Graves of EKOS Research investigates.


Frank Graves

Looking back at the Ontario election, we were very pleased to have predicted not only the victor, but also the majority. Only one other firm made this correct forecast. While we are very happy with the nearness of our final poll to the actual vote tallies for the two frontrunners, we were considerably less pleased with the performance of our Likely Voter (LV) model and our underestimate of NDP support. We were a bit over four points low on the NDP, but in a majority government this underestimate is of little practical consequence. The far more important challenges were identifying the winner, the leader of the opposition, and whether it was a majority or not; all of which we accomplished clearly and accurately. Yet in a closer race, the underestimate of NDP support could have been far more consequential and the challenge of forecasting who will vote and who won’t remains an important and unsolved puzzle.

As it turns out, the failure of the LV (LV) model, and the underestimate of the NDP are closely linked. The LV model significantly lowered NDP vote from a modest to large error and it moved the Liberal and Conservative votes from spot on to too high. Clearly, this wasn’t the intention and this error is somewhat baffling as the ingredients of the model are based on established historical patterns which show significant connections between these terms and likelihood of voting. To recap, we know that generally speaking (ceteris paribus), individuals who don’t know where there polling station is, who are younger and display lower socioeconomic status, who haven’t voted in recent elections and who don’t tell us that they are absolutely certain to vote, are less likely to vote. It is therefore extremely puzzling to see that applying these factors to estimate who will show up to vote not only wasn’t helpful, but magnified the error.

Further analysis of our final polls shows why this failure occurred and it is directly linked to the underestimate of NDP support. These could be quite separate issues, but an analysis shows they are interdependent. First, let’s look at the connection between our LV model and the final result.

The basic premise of our LV model involves assigning each respondent a “likely voter” score (maximum of eight points), based on a number of factors, such as age, past vote behaviour, knowledge of voting station, etc. Since we know that voter turnout is likely to be in the range of 50%, we isolate the (roughly) 50% of respondents who received the highest scores (giving us a “cut-off” of six points). Whether we apply the model weakly or strongly makes little difference. We find that different cut-off points make some impact but basically no model decreases error; they all magnify it. This is perplexing as we and others have shown clear linkages to these terms and voter turnout.

We then looked at the individual impacts of each of the terms of the model and none of the individual terms improved things. In part, the poor performance of the LV model is rooted in the paradoxical reverse linkages between turnout and the NDP vote. NDP vote appeared to be modestly rising in the later stages of the campaign, but we didn't make much of it because it was focused in parts of the voter spectrum that are not linked to high turnout: lower education, younger ages. 

Worse, the NDP vote was much more concentrated in non-voters from 2011, they were less likely to know where their polling station was located, they were considerably less certain to vote, and they were less likely to have voted in the advance polls. A few other notable findings: the NDP were performing more strongly with households with children at home and they were more likely to be union members. The link to the deep cuts to public sector workers may have caught the attention of these groups as the campaign matured.

One other interesting feature was evident in the data: NDP supporters were much more likely to only use a cellphone. We do sample cellphones, but we were under the population values and didn't weight this group up. This was because we were burned in the 2011 Federal Election for capturing cell only respondents who were much less likely to vote and less supportive of the Conservative government. In this instance, cellphone-only households were less likely to be Liberal or Progressive Conservative supporters, but they were not less likely to vote. If we had sampled more cellphone-only households, we would have probably been closer to the actual NDP vote.

In the end, we're left conclusion that even the most cautious and empirically informed attempts at creating LV models can be dangerous because things change; sometimes quite significantly. This harkens back to the classic problem of induction and the fact that the future will often not resemble the past. We are quite certain that presidential hopeful Mitt Romney was very surprised that he didn't win, as was Frank Newport of Gallup for precisely this reason. We therefore propose a more modest and less influential LV strategy that avoids the more ambitious approach we tried here. We also think that greater effort needs to be put into ensuring full coverage of all groups (e.g., cellphone-only households and younger voters), even if these groups hadn't been critical in past exercises.

We now have a reasonable handle on the reasons behind the underrepresentation of NDP supporters and the failure of the LV model. Our solution is a lighter less ambitious LV adjustment focusing on likelihood of voting. We also think that greater efforts to represent younger voters and cellphone-only households would be a prudent strategy.

We are, however, left with one big unanswered question. What is it in the latter stages of the campaign that seemed to engage what should have been a relatively disengaged NDP base to actually show up? That question is a very interesting one to which we have no real answer at this stage. A post-election survey would help to clarify this critical question.

What we do know is that this improbable appearance on Election Day was in defiance of historical patterns of likely voting and reinforces our conclusion to put more effort into creating the best random samples and less into seeking an elusive, unified field theory of voter turnout which we increasingly believe is a chimera.


  1. "less into seeking an elusive, unified field theory of voter turnout which we increasingly believe is a chimera."

    Says it all IMO

  2. Really enjoyed this, thank you very much.

  3. That was a really nice read. Being open about how their model worked, what its failures were and how they will proceed in the future was very interesting. I would love all pollsters to do the same kind of post mortem for every election.

  4. The really odd thing to me in the Ontario election was the fact that no pollster could come up with a reasonably accurate trend line.

    All of them really were basically calling for a tie and the results, except for Eric's last, simply did not get what was actually going to happen.

    So Frank Graves piece here helps but we really need an even more in depth study looking at all the pollsters and maybe figuring out why the results where so different from the polls ??

    1. Apart from two nights we never had the PC ahead and most nights had the stable lead we saw on Eday. None of the other pollsters had this trend which we thought was clear

    2. We need lots more studying and testing but it's hard to do with no sponsors. This does help us hone our products for paying clients though

  5. Nice post. Thanks for taking the time.

    Voter turnout in Ontario increased this election after decreasing in the five previous elections. That means that this election had a different dynamic than previous elections with regard to voter turnout. It could be something as innocuous as the weather.

    I would not throw out likely voter models just yet. For Ekos in particular, it corrected (overcorrected, actually) the problem of their consistent over prediction of Green Party support. Ekos has overpredicted Green party support for almost every election since I've been following polls.

    The Greens are a parking lot for protest votes. People who say they will vote Green may be less likely to vote, but it's obvious that some do. In the case of Ekos, they need a "likely voter" model to reapportion the Green vote, but it obviously needs to be done better than it has this election. It would be unwise for any pollster to abandon likely voter models just because version 1.0 needs a lot of work.

  6. It's mindboggling that between 87% and 94% of the electorate still has a landline in 2014. Landlines are singularly useless wastes of money.

    1. last stats can has it around 15 which we think is too low. Our best guess is that it now between 20 and 25% in Canada. Close to 50% in US (and their IVR can't call whereas we can under CRTC guidelines with strict standards)

    2. leonsp I wouldn't give up my landline even if they doubled the price. Better quality than cell phone. I have no interest in being "connected" 24/7

    3. If you have kids a landline makes sense still, although if you are single or in a no-kids situation it does seem like a waste.

    4. Gave up my landline years ago, the only people that called it were selling me something so couldn't see why I was paying for a conduit to bother me. Even though they can call cell phones people tend to screen them more, I know if I get an unknown number I often just let it go to voicemail.

      Liked the article but they down play their own importance. Their may have been some NDP support that went Liberal due to their under-estimating the NDP and over estimating the PC's. There was a strong Stop Hudak vote and when people use less than accurate polls this can have an impact. Consisdering this is only a 3-4 seat majority these poll errors could have been responsible to some extent.

    5. I have a landline as cellphone service in my neighbourhood sucks, but paying for 2 phones is a total waste of money. I'd love to ditch the cell, but can't.

    6. Cell reception often isn't great in basements. Some types of internet service are tied to landlines. Either one of these reasons would be sufficient for me to keep my landline, and both of them are true.

      That said, I almost never use a telephone of any sort. The spoken language is but a pale imitation of the written language.

  7. I think we need to accept that polls are not going to be able to 100% predict results all the time. What a likely voter model can give us is an idea of the range of possibilities when compared to the eligible voter numbers. So I don't think fine tuning to any specific election makes sense.

  8. And now we prepare for a flood of polls on Northern Gateway I think ?

    1. Yes, but, I don't think they'll have much impact on the situation. Most Aboriginal communities are opposed. The recent plebiscite in Kitimat was interesting roughly 57% opposed-such a result is not all that negative for Enbridge and demonstrates how almost evenly divided people are in Northern B.C. to this development.

      I don't believe Gateway will succeed. Aboriginal opposition and potential litigation is sure to tie up the project for years. By the time the project is ready to go the political scene may have changed in Alberta-that is how long the process may take.

      Most people on this site are probably too young to remember the MacKenzie valley natural gas pipeline; briefly, it was first proposed in the 1970's and construction began in 2010 on a route from Fort Simpson through to Southern Alberta. There are many reasons for the delay including low natural gas prices but, I think it safe to say the major delay was instrumented by Aboriginal opposition to the pipeline and route. In any case Gateway is far from a done deal!

    2. And something called the Dogwood Initiative as well which is going to run a provincial referendum and use its results to put a bill into the BC legislature to block Gateway. BC public seems to be against Gateway so like you I doubt it ever gets built.

    3. I'm as big a pro=-pipeline guy as you're likely to find, and I think Northern Gateway is the least likely - by quite a margin - of the big 4 pipeline proposals ever to be approved.

      Reversing the eastern link to Ontario seems like a no-brainer.

      The environmental impact of twinning the Transmountain pipeline is minimal (the right-of-way for construction already exists, and there has been tanker traffic in the Burrard Inlet carrying oil from the original Transmountain pipeline for over 50 years).

      Keystone will almost certainly be approved by whomever is President after Barack Obama.

      But I don't see how Northern Gateway ever gets built. It's like a red herring, intended to draw the ire of environmentalists while the other three carry oil.

    4. Ira,

      Hilary is not likely to approve Keystone XL. Frankly, its necessity is up in the air as accessible deposits increase due to fracking.

  9. The problem with polling formulas - likely voter or otherwise - is that phenomena shift. For the moment cell-phone-only voters are significant to the NDP's results, but in 4 years that margin is likely to evaporate... A formula may approximate results but it can't be predictive of them. My partner, who is a political analyst, from the start called this election a Liberal majority - based on a reading of hard-to-quantify but no less valid factors.

    As I've argued before, the combination of polls with deeper political analysis, is less fallible than either polls or analysis alone. Unfortunately, mainstream pundits do not apply deep (objective) analysis, only conventional wisdom, and they don't seem to be able to read polls (witness the number of misrepresented/misinterpreted poll results published every election).

    1. A good number of people have partisan blinders on, drank the cool-aid so to speak and really believe what they are saying. I fully assume the other group fall more into the category of knowing what they are doing and are just doing political spin.

      Judging by some posts here during the election you can easily see how that happens. Folks want so bad to be right or are so in the tank for a political party they are unable to read polling information.

    2. Carl,

      I feel your assessment is unfair. The accuracy of polls has rightly been put into question over the last number of years; the pollster mis-called BC and Alberta very badly. It is not simply about "reading polling information" it is also about trusting the information that is given. Pollsters still have a long way to go to re-gain that trust.

    3. I think there needs to be reasonable expectations, as well. People seem to expect laser-like accuracy from the polls, which is simply not possible.

      Even if the polls had been completely accurate, with the MOE they would have grouped together in a band of between 36% to 42% for the OLP, 28% to 34% for the PCs, and 21% to 27% for the NDP. A Liberal victory would have been the only real scenario considered likely, but not necessarily a majority. And gains or losses for the NDP would have been possible - all with the polls being technically right.

    4. Carl, it's true, there were some pretty outlandish predictions posted here (and elsewhere) for this election, and I guess the apparently contradictory polling encouraged that. Pete, your point about the apparent inaccuracy of the polls seems fair, and it just supports my point all the more... you can't rely on polls alone, you need to take the polls (with a grain of salt) and combine them with genuine political analysis. As a libertarian-socialist having grown up in Calgary, lived in semi-rural Ontario for nearly a decade and now in a Montreal riding that's been Liberal since the beginning of time, I've rarely had much inducement to think that my lesser-of-three-evils choice had much of a chance of winning (until 2011, anyway). It makes for a colder assessment of what's likely to happen, subject to whatever other errors I might make...

    5. Some people make accurate predictions, but that doesn't mean that those predictions were anything more than guesswork. They might be the result of some special insight, but we can't determine that based on a single prediction.

      At the start of the last provincial campaign in BC, my wife (a lifelong BC resident) insisted that the Liberals would win a majority, but Christy Clark would lose her seat. She said it was obvious, and didn't know why anyone believed the polls that predicted NDP victory, when it was so clear (to her) that they would lose.

      And they did lose. The Liberals won a majority, and Christy Clark lost her seat.

      I still don't know how she knew that.

    6. My point is that educated speculation isn't the same thing as blind guessing... and combined with a reasoned use of polling, it's possible to have a good sense of what might be happening. There will always be surprises, and one would be foolish to think there's a formula for predicting the future, but to come closer to what's actually going on politically is possible with some serious rational observation and analysis - which is often lacking in the mainstream media, for example.

    7. chimurenga,

      Libertarian-socialist seems to be an oxy-moron to me. How can one be in favour of freedom yet wish to confiscate and redistribute property?

      Also, your name "chimurenga" has a very negative connotation of late due to its association with Robert Mugabe. Although its original use was noble for some Mugabe has used the term to try and justify his various land grabs (not only from white farmers but, Blacks seen hostile to ZANU-PF) and re-distribution.

    8. The term 'libertarian-socialist' goes back to the middle of the 19th century. Libertarian-socialists oppose all concentrations power unless they can be justified in democratic terms and work to disperse power into the hands of all. Corporations, being inherently anti-democratic, and states, being democratic in only a limited sense (at best), are targets of libertarian-socialist critique and activism. Libertarian-socialists work to develop organisations of consensus-based voluntary associations (like collectives, cooperatives, syndicates/democratic unions, etc.). In North America the term 'libertarian' has been more recently appropriated to describe far-right, anti-collective, free-market capitalist individuals and organisations - the opposite of libertarian-socialism. Libertarian-socialists, being anti-authoritarian, also oppose the type of socialism you're referring to, as practiced in Communist countries in which collectivisation is forced (etc.), and instead support a concept and practice of socialism that is closer to the original understanding of the term: communal, mutually supportive, non-hierarchical, and democratic.

      Chimurenga means 'struggle'(in the Shona language) and refers to a long history of anti-colonialist, pro-human rights struggles in what is now called Zimbabwe. The term was particularly popularised by the musician Thomas Mapfumo (using it to describe his style of music) who was first an opponent of the white racist Rhodesian regime of Ian Smith, and then of its black-led successors, including that of Robert Mugabe. Mugabe has attempted to appropriate the term to support his political power, but in a cynical fashion that runs counter to the term's real meaning and history. Mugabe can't claim ownership of the word 'chimurenga' any more than Stalin could claim ownership of the word 'socialist' or Tony Blair can claim ownership of the word 'democratic'.

    9. chimurenga,

      Yes, I just thought you should be aware that your "name" has developed a very negative connotation in today's Zimbabwe. I understand Mugabe does not own the name but much like the "N word" or "Red Skins" have become offensive due to context and connotation "chimurenga" has a different meaning today than it had in the past.

      I think it a bit of a stretch to say "chimurenga" can be linked to pro-human rights struggles. Human Rights or even the concept or philosophy of human rights did not exist until post-World War Two. We should remember those African leaders who fought against European colonialism were fighting to preserve their own form of government and often their own privileges. The history of slavery on the West Coast of Africa certainly makes it difficult to understand how African leaders of the 19th and 20th century can be linked to human rights.

    10. I called myself Chimurenga, I hardly need a lesson on the meaning of the term. And equating "chimurenga" with "nigger" is nuts (and offensive). That's why I chose comparisons with similarly favourable terms, "socialist" and "democratic". Tyrants of all sorts lay claims to concepts like "revolutionary" and "democratic"; no one suggests that we should stop using the term "democratic" just because East Germany called itself the German Democratic Republic. Ask anyone of (black) Zimbabwean background whether "chimurenga" is a term they would no longer use... Your "history" of "chimurenga" and human rights simply doesn't accord with the facts. The word means "struggle" or (more loosely) "revolution", it was used to describe independence, anti-racist and anti-colonialist efforts from the beginning of the 20th century. Human rights are a feature of history from day one, under whatever name you want to use, they are a universal. Close to home, 18th and 19th century Eurpoean history is filled with examples of people fighting for their human rights (and in those terms). And your summary of African leaders is similarly out of touch. The independence movements of Africa sometimes were and have become profoundly compromised, but you can't tell me that Amilcar Cabral, Patrice Lumumba, etc. were just looking to preserve their "privileges" (whatever those might have been). Finally, how can you claim that the West African slave trade somehow fatally undermines 19th and 20th century African aspirations for human rights? By that logic, the United States has a much more feeble claim to support human rights, etc. because of its slaving past. Canada fares no better, what with our genocidal treatment of native peoples.

    11. One final note, the negative connotation that "chimurenga" has "developed" in today's Zimbabwe, is the same negative connotation it has always had for white defenders of Rhodesian apartheid. Black Zimbabweans do not share that view, regardless of their justified dislike of Mugabe.

    12. chimurenga,

      The "N word" originally meant black. Through the 17th and 18th centuries that was its only meaning in America. Its connotation changed however when it became a common derogatory term. Your moniker is going through a similar change at the moment. Where once it only meant "struggle" today it is used to justify Mugabe's racist policies and tyrannical rule. You acknowledge your moniker is offensive to some mainly, White Rhodesians. I am not asking you to change your name only pointing out that it is highly politicised moniker and at best controversial in the context of today. I think you may find however, that the term even among Black Zimbabweans is falling out of favour due to its association with Mugabe. Don't shoot the messenger.

      I am not debating the altruistic attributes of Lumumba. As he died so young we will never know how he would have developed as a national leader. However, the history of the Congo is rife with political intrigue and violent political machinations and there is no doubt some in what was the Belgian Congo used the power vacuum of independence to strengthen their own hand and preserve or enhance their power and privileges and commit various crimes including murder. Lumumba was naive as his untimely death unfortunately proves.

      I do not argue the slave trade undermines Africans aspirations for human rights. Merely point out human rights were not practiced in West Africa by the indigenous governments in the pre-colonial and often in the post-colonial eras.

      The US does have a weak claim to support human rights. Its system is based on citizenship. The US Declaration of Independence does not set out a list of human rights but "self-evident truths" based in natural law. Ironically it then goes on to limit those self-evident truths to white male citizens. Of course the propagandistic notion the US was founded on "freedom" or in order to give 18th century Americans "freedom" is at best open to debate. Natural Law is the term most often used in reference to what today we know as human rights in the period of the enlightenment until the 20th century.

      You are right Canada does not fare much better, our treatment of Aboriginal peoples is disgraceful. We should remember though that First Nations people before contact did not always act nobly or kindly towards other peoples. The Haida were known for their fighting skills, slavery was a common fate for a tribe or nation defeated in war. Pre-contact or Aboriginal on Aboriginal crime does not excuse European powers' actions acted toward Aboriginal peoples. It does however, lend attention to the nuances that traverse Canadian history and help give us some understanding as to why Europeans became drawn into conflicts. Undoubtedly, in the integration of neo-lithic peoples into the early-modern Western political and economic system that became Canada mistakes were made. I would not go so far to say it was outright genocide although certainly it is possible to see "genocidal attitudes" within government policy and actions. Today it is a difficult crime to prove since most of the perpetrators are dead and the ability to determine whether intent existed no longer possible in many cases. It is a question best left to history.

  10. Although this is a good interesting article, I regularly get annoyed when pollsters are proud of "we got the final type of government right" and discount the numbers being off. Their job is getting the numbers right - getting the government right is a consequence, not a result. A 2% error when two parties are tied could completely change the government, while a 5% error when one has a strong lead could be irrelevant, but I would prefer the former pollster.

  11. There is a new Forum national poll out today apparently showing the Liberals back up at 39 and the PC at 30.

    Eric ???

  12. True, that: http://poll.forumresearch.com/data/Fed%20Horserace%20News%20Release%20(2014.06.18)%20Forum%20Research.pdf

    39 LPC
    31 CPC
    19 NDP
    5 GPC
    4 BQ

    With my simulator, it would give me:

    154 LPC
    124 CPC
    59 NDP
    1 GPC

    And yes, that is a blank slate for the BQ. 17,5% in Québec seems to be the lowest they can go to actually get a seat, under it, they are wiped off the map.Obviously, the poll was done before the new leader was voted in and before he gets his "campaign" going, so that may change, byt for the moment, that is not encouraging news at all for the BQ.

    1. The BQ's new leader isn't going to help their fortunes much...

    2. Actually, it's their best chance at winning seats. People who didn't vote Bloc before aren'T going to vote Bloc ever, those who voted Bloc won't do so now if they present themselves as "defenders of all Quebeckers" as was shown in the last election and with the PQ, so going back to their roots and motivating the separatist base is the only way to go. Will it work, that's another question, but doing more of the same was an assured death sentence.

    3. For those interested, by region, it would go like this:

      24 LPC
      5 CPC
      3 NDP

      45 LPC
      24 NDP
      9 CPC

      67 LPC
      40 CPC
      14 NDP

      17 CPC
      6 NDP
      5 LPC

      31 CPC
      2 LPC
      1 NDP

      British Colombia:
      22 CPC
      11 NDP
      8 LPC

      3 LPC

    4. Thierry, I agree with your thoughts on the BQ. Motivating the separatist base is probably their best hope to retain and or gain seats in 2015. Separatism may be in decline but a significant minority will always seek independence from Canada. At this point in the BQ's history survival must be their goal and the best way to do that is motivate the base.

  13. Eric, from your latest update to the federal numbers...

    'The Liberals were up three points from Forum's last poll of May 22-23. This uptick has been attributed by Forum to the public having "forgiven" Justin Trudeau for his pro-choice stance, since his support went from 39% before the comments to 36% after and 39% now. Or, you know, it could be because those are all statistical wobbles within the margin of error that can be explained by something as simple as random chance.'

    Nicely said. But this is just the sort of thing that I find aggravating about the mainstream media (alright, in this case, a polling firm that is supported by and supports the mainstream media). Why would Canadians have to "forgive" Trudeau for saying something that nearly 80% of Canadians support? A fact that, being a polling firm, Forum must know. It's faulty reasoning based on political bias.

    1. chimurenga,

      80% of Canadians may support a woman's right to choose, however, that choice is useless if it is imposed. 80% of Canadians support women having a choice. Trudeau has not given his party members a choice he acted unilaterally. Therein lies the irony of his decision the Constitution of 1982 gives Canadians freedom of conscience and religion. Justin Trudeau has decided to revoke those rights for Liberals - now Liberals must choose between their conscience and or religion and their loyalty to the Liberal party.

      Trudeau needs to be "forgiven" because his actions have contradicted the principles of the Liberal party. Given the historical close association between Liberals and Catholicism I suspect his decision will cost the Grits votes in 2015.

    2. I don't think you;ll get much mileage out of that. A large majority of Canadian Catholics are pro-choice and, believing in the separation of church and state, want to see that freedom ensured in law. For the small number of people who are opposed to choice, I suppose this will be another reason to dislike the Liberals or Trudeau, but I can't see it being a significant factor.

    3. I am not an expert on canon law but, people who are pro-choice are not in communion with the Bishop of Rome. Priests have on a semi-regular basis denied communion to"Catholics" whose beliefs deviate from Church teachings and theology. As far as I know no cleric has been sanctioned for denying a sacrament to a known opponent of Church teachings. I would note as well that under canon law a Catholics who perform, assist or receive an abortion are automatically excommunicated. Does supporting pro-choice policies equate with assisting in an abortion under canon law? I don't know. In the most general description it may.

      Practicing Catholics make up significant populations in a number of ridings. Trudeau's stance may not be a significant factor on the next Liberal campaign but, certainly some will decide to support the church instead of Trudeau.

    4. There is a delightful irony if Catholics are upset with Trudeau imposing his and his party's viewpoints on those who want to represent that party, and excluding those who do not share a willingness to support that policy.

  14. And Thierry with those numbers the Liberals still don't achieve majority govt even in the current House.

    1. An 8% lead under 40% of popular vote shouldn't get you a majority. Well, in my mind, anything under 50% shouldn't get you a majority, but, if we apply FPTP logic, with three strong parties, 40% is generally speaking the barrier you have to pass, so it's not all that surprising.

    2. And wait till you hear the screams from the Conservative ranks over the Liberals "stealing" an election and getting an unfair advantage etc.

      Caramba !!

  15. Well, when you look at their "likely voter score" scale, it seems clear to me that the set of people who won't score high on it intersects with the set of people likely to vote NDP moreso than any other party. If you score someone's likelihood to vote low because of their age or income, and you then only count votes of the people in the top 50% of your scoring scale, it seems likely to me that you are not counting a ton of NDP votes. Also I would not worry about the "I don't know where my polling station is" question unless you are also asking "did you receive your voter card in the mail?" and the answer is NO. The voter card has the location of the polling station on it; all a person has to do is pick that thing up and they will know where to go... they don't have to have it memorized.
    Personally, as much as it might be nice to be able to predict the outcome of an election by using a likely voter model, I don't think that such a model should go much further than "are you likely to vote (yes/no)", because if you are intentionally weighting the opinions of the young and the non-wealthy as low, then that seems dangerous to me when you consider the potential impact that a poll can have on an election.
    More interesting to me, would be an inclusion of "does not support any party in this election / none of the above", which may actually be more likely to have a positive impact.


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.