Tuesday, March 14, 2017

John Horgan's NDP leads the B.C. polls, if you believe them


The provincial election in British Columbia is now less than two months away. And that means the polls are back.

After the failure of the polls in the 2013 vote, British Columbians might be forgiven for looking at them with skepticism. But the performance of the polls since 2013 suggests there is no reason to assume they are unreliable or that British Columbians are somehow unpollable.

Nevertheless, there is justification for looking at the latest numbers with a fair bit of caution.

You can read the rest of this article here.

19 comments:

  1. Voter turnout is always a problem for the New Democrats. The BC NDP has a very good ground-game to try and counteract this deficiency but, they are never entirely successful. The NDP ground game is particularly weak in rural constituencies of the Interior, if the NDP fail to win the Kamloops ridings it is unlikely they'll win the election.

    If, the BCNDP holds a 3.5% lead as the polling average suggests the Liberals would need to secure 57% of current undecided voters to nullify the BCNDP lead. However, the NDP vote is generally assumed to be more efficient (as evidenced by the 1996 Fudgeit-Budgeit-fast-ferries-Glen-Clark election). Generally, it is assumed the Liberals need to win the popular vote by 3% to secure a majoirity and overturn the BCNDP vote efficiency-that would equate to gaining roughly two thirds of the undecided votes, however, with the Conservatives unlikely to field more than a handful of candidates, the undecided vote is probably closer to 35% than 25%; if so, they only need to win about 57% of the undecided vote close to a third of whom are nominal Conservative supporters; advantage: Christy Clark.

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
  3. There's always so much vocal skepticism about polls when they show the NDP leading, but when they show the Liberals or Conservatives leading they always seem to be treated as gospel.

    It might just speak to the broader Canadian political mentality that expects the NDP to always lose. I've seen plenty of articles about how Alberta voters overwhelmingly expected the PCs to come out on top in 2015 right up until election day, even while poll after poll showed the ABNDP with a clear and sizable lead in voter preference.

    That expectation of defeat is something that all sections of the NDP (and the Greens) have to combat time and time again when it comes to voter enthusiasm and turnout -- and it's a handicap that the Liberals and Conservatives basically never have to overcome.

    ReplyDelete
    Replies
    1. It is not just the polls though, look at the UBC election stock market; it is at odds with the polls and while its record is not perfect it has usually performed better than polls.

      The BCNDP has won Government once in my lifetime. Bill Vander Zalm was almost universally unpopular and clearly used his office to further his own financial interests. Christy Clark is far from universally loved but, the same degree of animosity for her does not exist (except perhaps among the BC NDP).

      The reality is the best the NDP has achieved is 42.15% of the vote against an even more unpopular premier than Christy and they failed to win Government. 39% isn't an unreasonable number for them but, it's also not a winning number; Adrian Dix achieved 39% of the vote.

      Polls are great but, they need to be read accurately. The last two polls gave the NDP a four point lead but, they also come with a 2.13 and a 2.09 confidence interval. So, depending on how one reads the poll they either have a four point lead or are tied. We know riding distribution, boundaries and provincial demographics all favour the Liberals. We know the current polling numbers for the Conservatives are unlikely to materialise on election day as they will probably run far less than a full slate of candidates. We know the last time the BC NDP won Government their leader, Mike Harcourt was far better known than John Horgan is today and was up against a disgraced Government and a fractured centre-right.

      It's early days, campaigns matter aqnd the numbers will undoubtedly change. I just ask people take time to think what the nunbers acutually do mean, remember the margin of error and other externalities that favour or hinder a party's potential performance.

      Delete
    2. The corollary to your example above is that a four-percent polling lead for the BCNDP over the BC Liberals, combined with a +/- 2 percent margin of error, still means that the most likely outcome is a four-percent BCNDP lead, and that "tied for the lead" (NDP at -2%, Liberals at +2%) has the exact same likelihood or lack thereof as "eight percent NBNDP lead" (NDP at +2%, Liberals at -2%), both giving rise to very different results in terms of seats.

      But my point is, which interpretation do you hear presented more often?

      When it's the NDP (federal or provincial) with this sort of a slim lead, all you hear is "they could still be tied within the margin of error". Meanwhile, when the Liberals have a slim lead, they're "running away with re-election". Note the bias?

      As much as certain external factors you note above may tend to favour the BC Liberals in conventional wisdom, at the same time, the longer a party and a government have been in power, the more potential grows for a "throw them out" sentiment among the public (as we saw federally in 2015 and as we now see in Ontario). And even then, a solid campaign when the public are finally paying attention can overcome any and all of those factors -- and the election hasn't even been called yet. The public will know far more about all of their parties' leaders in the bright light of the campaign period when the media are actually focusing on every major group running for government.

      Delete
    3. Now I understand; you vote NDP.

      A friend of mine ran for the Dippers in Vancouver last federal election. Mid-campaign about half-way through August I dropped by for a quick chat. I said I thought the NDP in big trouble nationally and that it was going to be a tough fight for her. She brushed me off with polls claiming not only were the NDP going to win Government but, she was well on her way to winning her riding. Well we know what happened to prime minister Mulcair and my friend is back at her day job. It turned out the polls were conducted with suspect methodology without random sampling of the population-they were junk.

      I am not saying I am right. So, you don't need to take it personally. It's only politics! To continually put blind faith in polls when we know through past experience they tend to over-estimate NDP support is well...........The reason the BC NDP lost the 2013 election.

      Delete
  4. The polls say the NDP is ahead but, Liberals don't seem at all worried. Campaigns matter of course, however, most British Columbians assume the Liberals will be re-elected.

    Despite what the taxpayer funded BC Government ads say about B.C. having Canada's fastest growing economy and other economic platitudes for Christy everyone knows; B.C.'s economy is being boosted and is growing almost exclusively due to foreign investment in real estate from China. The economy isn't weak so much as fragile. A tariff or failure to re-negotiate the softwood lumber agreement with the U.S.would put the B.C. Interior, already suffering from lower Alberta tourist numbers, in recession. Pipelines and other natural resource extraction infrastructure (Site C) are needed to really get the entire B.C. economy humming not simply the Lower Mainland and Victoria. The NDP is the party of "NO" caught in an internal struggle between union workers and environmentalist they are unable to cobble together any policy with the ability to balance these conflicting themes. Instead they oppose the Kinder Morgan pipeline, even though Metro Vancouver needs more natural gas for its current domestic usages. Are against Site C and propose to take a huge loss on the mega-project even after many billions have already been spent and Site C is an essential element in B.C. Hydro's future plans and ability to provide electricity for B.C.; Whereas, Christy Clark is pro-business, pro-resource development and pro-Site C. Add to that John Horgan's relative anonymity and as of yet no campaign promises that inspire the imagination with an electorate and electoral system more favourable to centre-right parties.

    I am not saying the polls are wrong only they should be read cautiously. The NDP is ahead by 3.5% but, we know historically NDP voters are less likely to vote; whereas, Liberal voters being older are more likely to turn out. So what does a 3.5% NDP lead really mean?

    People don't love Christy Clark but, the BC Liberals have remained relatively free of scandal and compared to their Ontario cousins have shown themselves to be competent, if somewhat miserly, managers of the Province and economy.

    Eric's prediction of a NDP majority seems to me optimistic at the moment.

    ReplyDelete
    Replies
    1. If assumptions of victory were what won elections, then the Alberta PCs would still have been in power after May 2015. Everyone assumed they were going to win even as the polls increasingly said otherwise, just because the conventional teachings about Canadian politics were so deeply ingrained.

      To Eric's substantial credit, he keeps as much of his own subjective interpretations to a minimum as he possibly can and lets the numbers speak for themselves. Now, the pollsters, on the other hand -- their various formulae are proprietary, so we may never truly know.

      Delete
    2. The UBC B.C. election stock market is data. Yes, it is a compilation of people's perceptions but, it also is a community of equal or greater numbers than most provincial polls. While, I would not claim all investors are "experts" I would think a level of familiarity with British Columbian politics inhabits many if not most of them. They use their own money with the intention of turning a profit: people may lie to pollsters on occasion either because they don't wish their preference known or for some other reason. No incentive to be dishonest exists with the UBCBCESM and its track record speaks for itself.

      The discrepancy currently between the UBCBCESM and Eric's prediction does deserve examination. Eric's predictions are only as good as the polls he uses to construct seat predictions and through them electoral outcomes: majority or minority government. The UBCBCESM uses human intuition based on knowledge from a variety of sources, including but not limited to polling data. Neither is necessarily better than the other: Eric's methodology is exclusively based on quantified data; whereas, the UBCBCESM, depending on the investor, may be based more on qualitative data or quantified data, however, its results; share prices; are quantified data, they are investors' willingness to pay for shares on the popular vote market, seat share market or majority government market.

      Currently the UBCBCESM predicts a minority Government with the BCNDP or Liberals winning a plurality and the Greens holding the balance of power. However, in terms of majority Government the Liberals are seen as more capable of doing so with a 60% chance compared to a 38% probability for the NDP and a 6.5% chance of a hung Parliament. By contrast Eric predicts the BCNDP currently have an 85% probability of securing a majority government.

      The NDP could win a majority government in seven weeks but, that is not where the race stands at the moment and it is the numbers that tell us so; the polls inform us either the NDP have either a statistically insignificant 4% lead or it is a tied race. The UBC BC election stock market tells us it is a tied race.

      So you can disregard half the data if you want and discount the 15% probability just like Hilary Clinton or one can be curious and examine the numbers in full for what they are and recognise probability is not the same as certainty: externalities exist that all things being equal are likely to favour the incumbent.

      You keep bring up the 2015 Alberta election, Eric did well in his prediction but, the NDP received 4% less than their polling average! To think the current polls are overestimating BCNDP support is reasonable, cautious and most likely correct. It happened in 2013 when the BCNDP only took 86% of their polling average and it happened in the 2015 federal election as well. We know NDP supporters are less likely to turn up to the polls than their Tory or Grit contemporaries. It is why the BC NDP has developed a strong ground game and why during some campaigns they have gone house-to-house in certain Vancouver neighbourhoods to get out the vote! It is a longstanding problem for them for which you are pleading willful ignorance.

      Delete
    3. I would like to point out that the UBC stock market followed the polls nearly exactly in 2013, doing no better.

      https://predictionmarkets.ca/BC13pred.php

      So I'm reluctant to put too much stock in the results, though they are interesting.

      Delete
    4. I would also point out that the probability estimations assume the election is being held today, based on the polling information currently available.

      Those placing bets in the market are predicting the outcome in May. If I were to do the same, I'd cut the Conservative vote and boost the Liberals.

      That will eventually happen in the polls anyway and the projection will cut the Conservative vote down once their slate of candidates is known.

      Delete
    5. So, I'm being told not to put blind faith in polls while you (i.e. "WGS") put blind faith an "election stock market" and in the Liberals always over-performing and the NDP always under-performing. Right. That's so totally convincing.

      Of course the various sections of the NDP tend to have problems with turnout, enthusiasm and softness in voter intentions. Everyone else the whole country over is incessantly telling us how it's supposedly not worth voting for our party of choice "because they'll never win". That doesn't deter me, but I'm sure it weighs on a lot of otherwise leftist/progressive voters.

      For my part, I'm looking forward to seeing what the polls do when the election has finally been called and average citizens are truly paying attention. Then the spotlight will be on, the votes will no longer be parked, and the parties'/leaders' actions will have impacts on voting intentions.

      Delete
    6. Eric,

      I am afraid to tell you, you are not correct. The stock market performed better than 308 on the popular vote market. Discrepancy between NDP result and prediction 308: +4.5%; UBC +4%.
      BCL; 308:-7.94%; UBC -7.74%. Greens: 308: +1.5%; UBC +2.8%. Conservative; 308: +2.15% UBC: -1.25%

      Total deviation: 308 16.09%
      UBC Election SM: 15.79%

      So, the UBC stock market did perform slightly better but, both were off considerably due to "faulty" polls or perhaps "faulty" pollsters or pundits.

      Delete
    7. I am not telling you anything!~ Why are you taking this personally? My goodness, thin skin is one thing but, pooh hooing and moaning because other people have different viewpoints quite another. All you say is that you have a closed mind and are unwilling to accept data that conflict with your opinions or preconceived world view. Good job!

      The last Alberta general election polls overestimated NDP support by 4%. The last six polls from the 2015 federal election over-estimated NDP support by 1.4%, the last BC election the poll averages over-estimated NDP support by 7.94%. Do you see the trend? Polls tend to over-estimated NDP support. The 2011, 2008 and 2006 federal elections final round of polls also over-estimated. Given, the discrepancies were not nearly as large as 2013 in B.C. but, the trend is clear and the conclusion is also: NDP and Green supporters have a lower turn-out rate than Liberal or Conservative supporters. The NDP has known this for well over a decade! They have developed a very good ground game in B.C. to try and mitigate the effect and have spent considerable time, money and effort to do so!

      How do I know? I volunteered and voted for the BCNDP for a short time in the early 2000's when I lived in Vancouver-Fairview. The campiagn I was a part of well knew the poor turnout of their potential voters. We spent most of the nomination campaign and all of the general election campaign planning and organising for election day with car-rides, ensuring we had high advnace poll turnout and the like. We won by 1,000 votes

      A.S., you are denying a problem the NDP has already accepted. The data proves NDP turn out is often lower than its poll support and we know there are a host of corollary socio-econmic data on income, education, efficacy etc... that verifies this conclusion.

      All I asked: what does a 3.5% NDP lead mean when we know from history turnout favours the Liberals and, we know demographics favour the Liberals and, we know incumbency favours the Liberals and, we know riding distribution favours the Liberals? So what does that 3.5% really mean? How does 3.5% translate into the Legislative Assembly?

      A.S. you seem intent to deny me the opportunity to ask such questions but, we only need look at how "well" the polls did last November or in May 2013 or last June 23rd to see their fallibility.

      You are quite right the longevity of Government and the personality of the premier are important factors. The B.C. Liberals are long in the tooth. By the numbers of historical averages they should lose this election. The polls favour the NDP but, a 85.5% probability of the NDP winning the election seems optimistic; especially with seven weeks left until election day when other data favours the Liberals.

      Delete
    8. WGS — a better performance of 0.3 points across four parties is not something I'm very concerned about. What I am concerned about is markets claiming they can know better than the polls, when the 2013 BC election demonstrated they are based primarily on the polls in the first place.

      Delete
    9. Eric,

      I think that is a very cynical analysis.

      If one looked at the whole history the market does perform better on average than do polls. Not by any great amount but better. Of course the information is primarily based on polls but, when polls have proven unreliable of late, one should remember the better record even if it is only slight.

      The pollsters were wrong in the U.K. because of turnout. Pollsters were wrong in the U.S.A. because they failed to accurately monitor or calculate turnout. And pollsters were wrong in British Columbia in 2013 because they failed to account for turnout. So, if one is not going to adjust their predictions or estimations based on turnout-they are better to use the UBCBCESM instead of polls for data because we know historically the UBCESM performs slightly better at least compared to the final prediction or last round of polls.

      Delete
    10. Again, my point is that if the market missed the B.C. election in 2013 in the same way as the polls did (even if in that case it might have been a few, relatively meaningless, tenths of a percentage point closer) than it does not contain relevatory information or insight that is not available to pollsters.

      If the market was able to tap into something pollsters cannot, then it shouldn't have replicated the general narrative of the polls in the 2013 B.C. election.

      It is still an interesting tool to follow but it is no replacement for polls.

      Delete
    11. Well, the UBC ESM is no replacement for polls so long as polls are accurate. We can agree on that.

      Also, I hope you have tweaked your methodology for B.C. In 2013 you made some riding predictions that in hindsight (which of course is 20/20), were strikingly off, particularly in suburban Vancouver where riding demographics have changed considerably over the last twenty years. This was most notable in the "win probabilities". For example; Surrey-Fleetwood had an 89% projected probability of a NDP win, it barely went Liberal by less than 1%, however, considering the federal riding was held by the Tories at the time and had been for the better part of the previous decade an 89% probability of an NDP win, even with the "faulty" polls, is circumspect. The NDP still should have been favoured albeit with a lower spread.

      Delete
    12. I am not aware of any evidence of a particular overestimation of BC NDP support in polls, outside of the 2013 case. The Liberals generally led the NDP, outside the margin of error, in polls conducted during both the 2005 and 2009 campaigns -- elections which the Liberals won somewhat more narrowly than in 2013. Indeed, as Eric's blog article of Mar. 21, 2013
      points out, over the 1996, 2001, 2005 and 2009 elections, polls leading up to the election have tended to
      underestimate NDP support, and overestimate Liberal support. The 2013 election was conventional enough in its outcome that I wouldn't think it a game changer in these regards, though living in Ontario, I have no particular knowledge of local factors/issues or evolving demographic conditions in BC.

      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.