Tuesday, March 31, 2015

Forum tried to poll PC party members. What happened next was strange.

Polling for a party's leadership campaign is fraught with hazard. While it can reveal how the general population and party supporters feel about the leadership candidates - and that is useful information - they can't tell us anything specific about the actual outcome. Only a very small proportion of party supporters actually vote in a leadership race, and that proportion is even smaller when we are talking about the general population.

There are two ways to go about polling a party's membership. One way is to obtain a list of party members and call them, which has happened in the past. The less common route is to do a standard poll and ask people if they are party members.

That's what Forum did in its latest poll concerning the Ontario Progressive Conservative leadership race. Good for them for trying, but logistically it should be a nightmare. According to the PCs, they have over 80,000 members, representing 0.6% of Ontario's adult population. You'd have to do a poll of 50,000 people to get a decent sample size of 300 party members, if the poll is done properly. With the response rates that IVR gets, you'd have to call about 2.5 million households.

Instead, Forum polled 881 Ontarians and found 65 paid-up party members. This is revealing, for several reasons.

Those 65 paid-up party members represent 7.4% of the sample, quite a bit more than what it should have been if the poll was completely random. Forum found 12 times as many PC party members in its poll than it should have.

This raises a few issues. First is the question of whether we can believe these 65 respondents. If the poll was done correctly, some of them would have been lying or misunderstood the question.

Second is the question of accuracy. If Forum could find 12 times as many PC members as it should have, what does that say about its sampling methods?

The third question is, perhaps, the most interesting one. What if active political partisans are just so much more likely to participate in a poll than the rest of the population, that being over-represented by a factor of 12 is actually not a sign of a problem with the poll, but a problem with the people who take part in a poll in the first place?

Response rates are tiny these days, and particularly with IVR polling. Nova Scotia's electoral laws require pollsters to report their response rates, and when Forum did in 2013 it showed that rate to be about 2% to 3%. That is a large proportion of potential voters not bothering to pick up the phone or complete the survey.

If that 2% to 3% is largely made up of people who are active partisans, does that skew the poll? Other polling has shown that poll-takers are more likely to be politically active and do volunteer work. The important question, then, is whether these people are politically different than the rest of the population. If they are, than the poll will be led astray. If they aren't, than the poll will be effective.

Putting aside these existential questions, the poll itself had a counter-intuitive result. PC party members broadly lined-up with the views of the general population and of PC supporters.

Of those 65 people who claimed to be party members, 51% said they preferred Christine Elliott. Another 13% said they preferred Monte McNaughton, while 10% opted for Patrick Brown. In the chart above, each box represents one of those 65 respondents.

Another 14% said none of these three contestants were their preference, while 11% were undecided.

Among PC supporters, a larger group, Elliott had 35% to 15% for McNaughton and 11% for Brown. Among all Ontarians, Elliott had 24% to 7% for McNaughton and 6% for Brown.

There are many caveats with these sorts of numbers, some of which were spelled out above. But the most important is, of course, sample size because the margin of error of a sample of 65 people is +/- 12.2%.

When we look at the poll with these margins of error, we get a much less clear picture.

(This is also a good example of how the margin of error applies differently depending on each candidate's support. Brown, at 10%, cannot be as low as -2%.)

Now we see Elliott at between 39% and 63%, McNaughton between 5% and 21%, and Brown between 3% and 17%.

If we look at just the decided PC party members, the margin of error increases to +/- 14.1%, and Elliott's support ranges between 55% and 83%, compared to 7% to 29% for McNaughton and 4% to 24% for Brown.

By these numbers, we would consider Elliott a lock to win on the first ballot. McNaughton and Brown are fighting for second place. However, we also have a pretty good idea that this is not the actual contours of the race.

We won't know the final tally of each candidate's membership drive until the end of April, when the PCs release the numbers. But Brown has claimed about 40,000 members, with Elliott claiming 34,000 and McNaughton 20,000. One oft-quoted source puts McNaughton closer to 6,000, which is a more intuitive result. We just don't know for sure, though any misleading candidate will be embarrassed once the numbers are announced.

So the numbers are probably not too far off, particularly if McNaughton is at 6,000 instead of 20,000 (the three claims add up to 94,000, but 80,000 if the source is correct).

At those rates, Brown should be somewhere between 43% and 50% of newly signed-up members, compared to 36% to 43% for Elliott and 8% to 21% for McNaughton.

This ignores the voters that were already with the party when the race started (some 10,000) and those signed up by Vic Fedeli and Lisa MacLeod (both of whom endorsed Elliott).

So now we get to the question of accuracy. Considering the limitations of Forum's poll, their Elliott numbers are, perhaps, plausible. Those undecideds could be from the ranks of the original party members, who should represent some 13% of the total. But to have Brown and McNaughton within the same error bands with at most half of Elliott's support, and McNaughton narrowly favoured among the two, does not quite align with what we know about the race.

This really puts the poll in question, and highlights the limitations of this attempt at getting at the voting intentions of party members. Does Elliott really have the kind of advantage that Forum gave her? It doesn't seem so, but we won't know for sure until the votes are counted in May.


  1. IMHO Forum really embarrassed themselves bye even reporting on those numbers fro "party members" - but they stand out among polling companies for having no shame!

  2. That political partisans are more likely to respond to polls wouldn't surprise me at all. I follow politics closely, and I always respond to polls. I was on Angus Reid's online panel. I've been polled by EKOS several times. Even Nanos called me.

    I respond because I want my voice heard. I know what polls can do to a race. So yes, in my eagerness I skew the results.

    1. This....I suspect this is a problem with all polling companies and not just Forum...Forum at least admits it here why their job in creating a high quality poll is so difficult.

    2. Any self-selecting group (those that answer the phone) are going to be predisposed to the select criteria. However, that same or similar selection process occurs with the people that actually vote. I think this is why in 2011 "gross numbers" were more accurate than "most like to vote" numbers. The adjustment was already included in the polling method, thus specific tweaking resulting in overestimating the level of partisanship.

  3. This comment has been removed by the author.

  4. There should always be poll validation question:

    Are you a union member?

    There are 4,735,367 union workers in Canada There are 23,971,740 people eligible to vote.

    If a sample does not have 19.8% acknowledged union membership within the Confidence interval it should have to publish a disclaimer that the sample is not valid.

    If I was paying for the poll I would demand it.

    If I was a news organization and wanted to protect my reputation I would demand it before publishing the poll results.

    Failure to take such a simple validation step calls into question the integrity of any news agency publishing polls,

    This disclosure by Forum in this poll is almost a plea for mercy. Don't believe any poll we publish.... we just can't get a valid sample.

    1. The problem with asking people a question like that is you are trying to subdivide the populace to get the most representative sample. Normally I would agree with that, but there are a billion and a half way to subdivide the population.

      On the other hand I agree Forum is basically saying we can't trust public polling, especially their own ... not that their polling over the last two years has given people any reason to trust them.

    2. The validation question does not have to be used to slice or dice the data... Just make sure that on some level the sample is a reflection of the known universe that it is purporting to report on,

      The media published polls have been taking invalid samples and then making adjustments to better reflect the known voting population.

      another validation question is as simple as age.. It is a known that 25% of the voting population in Canada is between 18-34. If a sample does not get 25% +/- the confidence interval it is not a good sample.

      The samples are consistently so far off that the polling companies have formulas to adjust the demographic inaccuracies of their polls to reflect the know reality.

      That is not good enough... If you can't even get the age sampled correctly then how accurate is the political preference???

      If you have a sample of 3000 as does EKOs it should get the age demographics correct within 1.8 pp.. ie if they have the sample of 18-34 anything different than 23.2-26.8 then their sample is invalid and should be labelled as such.

    3. Your suggestion that people be asked if they are union members is spurious. This really ought to be a forum for discussing polls, polling and related issues, rather than a soapbox for disseminating highly personal messages. I am not going to engage you on the politics, but just two the practical aspects of your suggestion. The purpose of asking questions is to validate the makeup of the sample, and the questions reflect (typically) demographic generalities and specific partisan leanings as a check upon expected sample rates. Your first problem is that (based on partisan political beliefs) you believe that union membership is congruent with a set of partisan beliefs. That is simply, and demonstrably, false. In fact, the largest grouping of union members tends to be entirely apathetic, without consideration for any specific policies or parties and how they may affect unions. Another large group is union members who frankly vote against union interests for a variety of reasons. There is a third group who do, in fact, vote according to what might best serve the union, but it is a minority. So the first problem is that the question serves no purpose, as union membership does not correlate with partisan support.

      The second problem is that, by suggesting pollsters identify samples and issue disclaimers based on broad, non-representative categories, you open the discussion to a chaotic miasma of irrelevant possibilities. Replace "unions" in your post with "teachers", "bankers", "oil company employees", "civil servants", "regular churchgoers" and you begin to see how unworkable and undesirable this would be. And frankly, it would serve no purpose as far as I can tell...

      ...Other than, perhaps, to try and advance some kind of partisan message? That polls are to not to be trusted if we don't like aspects of the sample which have no direct bearing on how people vote? Honest, I don't understand how this advances anything.

    4. BCVoR,

      Your objections could be overcome by a quota system or throwing out oversampled datasets. Let's say that typically 18-34 are oversampled and 65+ are undersampled. You get a smaller overall dataset, but it will conform to your desires. Isn't more information always better, even if you have to make adjustments between polling statistics and population statistics?

    5. Mapeleson

      NO It shows that you are not doing a valid survey.

      your results are not statistically valid.

      whatever sampling technique you use somehow catches more 18-34 year olds ... it would be a totally invalid assumption that the 65+ that you have caught in your sample are representative of the 65+ group as whole.

      You end up with a poll where your sampling technique is biased to a certain demographic group of people it is not random and the results are not valid. You are not entitled to call it a scientific poll or say that it is accurate within a certain range 19 time out of 20.

      take it to the extreme and say you can only get 10 people over 65 and they all say they vote Green.

      It ends up that these 10 peoples views have to be extrapolated to represent 20% of the voting population. Green with 10 votes out of 3000 gets a 20% The MOE would have to be calculated on a sample of 10

      The Green have 20% support +/- 25%.

    6. It seemed I touched a nerve...

      The reason I suggested union member because it is a large portion of the candain electorate that we know the exact number with a great degree of accuracy.

      The same could be said for Teachers or doctors or nurses. We do not have any accurate count of church-goers.

      Do you agree that the raw number of age (a question that is already asked in almost all surveys) should be used to validate the sampling.

      When EKOS does a survey sample of 3000 people...Using ekos because they do large surveys they break down the demographics by age.

      35-49 years old --- we know that there are 7.22 million Canadians in this age category accounting for 26% of the possible voters.

      If in your sample of 3000 you get less than 768 or more than 796 your sample is the 1 in 20 out;lier

    7. BCVoR,

      Validity is a matter of degree, not binary options. What I was saying is that polling techniques can be used to achieve your given result (specific subgroup representation), but it doesn't actually improve the validity of the poll.

      The basic premise of all polling is that a "random" sample is representative of the whole. The reality is that any poll (beyond the old mandatory census long-forms) is a self-selecting group, and therefore not truly random. This doesn't matter if you are using the internet, telephone, or talking to people on the street (although each has it's own biases).

      The true question is: given the bias in sample selection, how well does this translate into actual results, and how accurate are any adjustments to counter this inate bias?

      Politically polling uses demographic data to adjust the weighting of polls because this has historically translated well into representation of reality. Thus, there will be greater variability in the breakdown of 65+ doing internet polling than in 18-34 year olds, but it does not mean the randomness of the group is significantly different. If there was a bias to say NDP voters not having internet access, then the poll suffers validity.

      The chances of getting a sample of 10 people 65+ out of 3000 who all vote Green are astronomically small. Let's give the Greens 10% actual support, thus 10 randomly selected seniors all supporting the Greens is one billionth of a percent. It is much more likely to get 0 Green supporters (34.9%) that that level. Thus, your scenario falls outside the confidence interval and represents the 20th time out of 20.

    8. BCVoR,

      Being outside this range doesn't invalidate the whole poll, it just adds to the margin of error. Rather than an aggregate error, it should be calculated based upon the accuracy of each subgroup.

    9. That reminds me of Health Canada's assertion that the effectiveness of this year's flu vaccine was somewhere between 23% and -4%.

      So that would be 9.5% +/- 13.5%.

    10. BCVoR, if union workers should be identified, then why not teacher, doctors, accountants, company owners or minimum wage workers? Or, if you want a group with a likely bias, oil company workers? Simple answer, probably because they don't relate very well with voting intentions and it would be a huge pain to also take that into account. Polls are meant to give an accurate representation of the population, up to a certain point. There is always a "bias" that is included in the form of what criterias are selected for the representation (the same can be said of pretty much any study). In political polls, it's age as it greatly correlates to voting patterns. How do union workers vote? 5% of them? 50% of them? 90% of them? How do you adjust that into voting intentions, as opposed to age? If selecting a random group that has a bias against union workers, it will simply introduce another bias, skewed towards another group, whichever it is. In the end, the number of union workers is probably close to what it is in real life, and if not, does it really change anything? Do union workers really vote vastly differently to the rest of the population?

      And while we're at it, why stop with union workers, minimum wage workers, doctors, teachers, accountants or company owners? We could include retail workers, lawyers, political activists, greenpeace members, gun owners, farm owners, psychologists, pharmacists, publicists or construction workers! You know how such a representative sample is called? A general election.

    11. While I believe you're right, the negative percentage could relate to induced cases, where individuals became sick when they otherwise would not have.

    12. Mapleson

      The nanos survey I analysed by age demographics was out by 22% of sampled 40-49 people.

      19 times out of 20 it is supposed to be out by 3.1%

      If the random sample is so verifiable wrong how much confidence do we have in the unverifiable question ?

      The only straight question Nanos asks in the poll (outside of demographic info - age, sex,province) is "Question:Of the current federal political party leaders, could you please rank your top two current local preferences for Prime Minister? [ROTATE PARTY LEADERS]"

      Only the top ranked choice is reported.

      so with the proven accuracy of the sample of +/- 22 pp

      Harper is the preferred choice of between 9 and 53% of Canadian voters and Trudeau is the preferred choice between 8 and 52 %.

      Apparently polling companies do not have the ability to put together statistically valid samples.

      The universe has become too diverse to put together a statistically valid sample.

      It will continue to degrade as survey participation rate trends to zero and/or they are high jacked by political operatives.

      Any poll they publish is basically a commentary rather than statistical valid reflection of the current situation. Ironic in that most all Canadian pollsters also double as commentators.

      This is the reason that some pollsters have moved to panels and using Baysean Inference instead of the more traditional Normalized distribution survey sampling. If you can't put together a randomized sample you just can't trust the results.

  5. Eric, a new poll from Mainstreet Technologies with a very large sample size (3,000) has been released showing support in the Alberta provincial landscape with interesting results. The PCs and Wildrose were tied provincewide, while the NDP had a huge lead in Edmonton. I'm sure you've seen it by now, and am interested to see how the poll would translate into seats based on your model.


    1. See the replies in my Twitter feed.

    2. The WR Leader has done a great job of not looking crazy recently, and the defeat of Danielle Smith in her PC nomination meeting neatly closes the book on that chapter of the party's history.

      It would be exceedingly funny if the leadership abandoned the party only to see it win. They should have learned from the Reform Party - they polled exceptionally poorly between elections, but did well when votes were cast. The Moribund polling of the WR between elections wasn't meaningful - it was just the predictable bump for the new Premier.

    3. I saw one serat projection from Blunt Objects Blog (whose authors are admittedly biased toward all Canadian Liberal parties) that had PC 36, ER 24, NDP 15, Libs 11, AP1. I tend not to take these guys too seriously because: First they are biased and secondly, in my opinion they are amateurish. As you can see by their seat projection they have only accounted for 82 seats in a House of 85 seats. That is a mistake that really should be avoided by any self-respecting journalist or blogger or even a good proof reader.

  6. Éric,

    Assuming for a moment that Forum is admitting there is a problems with their polling, how would it effect the model?

    What would be the results without Forum's data added into the model?

    Going one step further, if we assume all public polling is effected by the same problems as outlined by Forum, is there a way to compensate for it? is it just mater of increasing the high/low and min/max ranges?

    1. The mix/max ranges are based on how polls performed compared to actual election results, so they take into account past errors.

  7. Eric,

    firstly, I believe a new poll on Alberta politics is out from Mainstreet which I hope you'll be able to write on sometime this week.

    Secondly, I am a little confused by your comment that: "Only a very small proportion of party supporters actually vote in a leadership race".

    While I can see this to be true of the traditional delegated convention in the one member one vote system I would be surprised if this was the case. For instance the LPC claim that 82% of eligible voters cast a ballot in the 2013 leadership contest. Off the top of my head I would think that in a one member one vote system turnout would at least be equivalent to the public at a general election-50%. Am I missing something?


    1. When I said party supporters, I meant party supporters - not just people eligible to vote in a leadership race. The vast majority of people who will cast a ballot for the Liberals in 2015 did not vote in the 2013 leadership race.

  8. Polling changes from week to week, people change from week to week, The Polling firms have a tricky assignment to measure what would be accurate.
    I think they do a great job contributing their findings on a regular basis.
    I especially appreciate how 308.com attempts to combine results.

  9. Nanos weekly tracking report published April 1, 2015 is not a valid sample.

    The sampling technique would give the wrong answer to the number of people in Canada between 40-49 being out by 22%.

    Known facts:

    Canadian by age and percentage of voting universe
    18-29 5,765,380 20.2%
    30-39 4,847,000 17.0%
    40-49 4,850,800 17.0%
    50-59 5,331,900 18.7%
    60+ 7,753,400 27.2%

    Nanos give the breakdown of his sample by age in his report:
    18-29 205
    30-39 169
    40-49 208
    50-59 178
    60+ 239

    With the sample size of 1000 this should be accurate within 3.1 pp giving the following high low ranges that we would expect 19 times out of 20. The perfect sample is what the survey should include with knowledge of the universe.

    18-29 199 211 perfect sample 202 (inside the range)

    30-39 164 174 perfect sample 170 (inside the range)

    40-49 202 214 perfect sample 170 (way over sampled out by 22%)

    50-59 172 184 perfect sample 187 (outside of range under sampled by 5%)

    60+ 232 246 perfect sample 272 ( way under sampled by 12%)

    1. Using your reasoning, BCVoR, it's basically impossible ever to have a valid sample, so... what? No polls should ever be published?

    2. BCVoR,

      You continue your abuse of statistics.

      First, population data are estimates, not facts. They are relatively stable and accurate as we have estimates of the death rates for each group plus the annual categorical line hopping and immigration (emigration data is less reliable). For the 2011 data which these 2014 figures are based, the Margin of Error is 0.5%.

      Second, only 999 people are included in the age totals.

      Third, if you drill down you can get better data from StatsCan.

      18-29 5,831,378 20.4%
      30-39 4,846,997 16.9%
      40-49 4,850,840 17.0%
      50-59 5,331,943 18.6%
      60+ 7,753,387 27.1%

      However, this is the July 1, 2014 estimation, and the poll is taken in March, 2015, so approximately three quarters of people will have crossed the line. From the data:

      17 422,155
      29 495,415
      39 470,625
      49 542,849
      59 489,202

      Thus, ignoring deaths and immigration, the numbers shift to be:

      18-29 5,758,118.00 19.8%
      30-39 4,871,787.00 16.8%
      40-49 4,778,616.00 16.5%
      50-59 5,385,590.00 18.5%
      60+ 8,242,589.00 28.4%

      For 2013-14, total deaths were estimated to be 256,721, immigrants were 267,716, emigrants 61,928, and returning emigrants were 36,811. I can't seem to death data beyond 2011, so here are the July 1, 2015 StatsCan population forecasts:

      18-29 5,820,900 20.1%
      30-39 4,899,400 17.0%
      40-49 4,799,300 16.6%
      50-59 5,371,200 18.6%
      60+ 8,008,600 27.7%

      All of this goes to show that while population numbers are relatively accurate statistics, they aren't facts. In each of these cases, the number of 40-49 are equal or lower to your assumption, so the issue goes beyond assuming estimates are real. However, Nanos' poll was not of all Canadians over 18, but it is a random telephone survey conducted with live agents, reaching out to Canadians through a land-and cell-line dual frame sample. Thus, no telephone means no chance of being included. Thus, the number of seniors decreases (to start those that live in retirement homes) as well as those under 30 (in an internet world, it's easy for the tech-savy to live phone-free).

      You can't just mash two random numbers together without considering their relative usefulness and accuracy.

    3. Mapleson....

      you worked hard with numbers.

      But you never got around to challenging that Nanos (or any other pollster) can not get a valid sample of the voting population...

      Last time I checked there is no requirement to have any phone in order to vote.

      It is just too hard to get the proper sample of people over 60 ... it doesn't make the sample correct to ignore them.

    4. Ira You are correct.. Polls should not be published unless they can prove the validity of their sample.

      If you can't put together a randomized sample you just can't trust the results. Pollsters can't get a valid random sample of the voting population.

      This is the reason that some pollsters have moved to panels and using Baysean Inference instead of the more traditional Normalized distribution survey sampling.

    5. So let's close this site down and just let pudits say their party is ahead because they said so? The results gained do over sample some categories and under sample other categories, when comparing to a fixed total, but the information gained in each category has some value so long as it fits historical context.

      Polling is more an art than a science.

    6. Its not like an over sampled bracket cannot be correct for.....

    7. Mapleson

      Great comment...

      Polls need to be published with caveats to say that they are not scientific.

      They are not going to be correct 19 times out of 20 and the margin of error is much greater than stated.

      It is not a matter of over-sampling.... it is not being able to set up a valid random sample.

      The historical context is not at all valid .... There are no facts on how any demographic group voted in the last or any election..... There is precise information on how many voted but the private ballot does not allow the breakdown to provide an historical context.

      Eric at 308 is just taking the information provided by these polls and attempting to average them out in a scientific manner to account for the various polling artists tastes.

      He is handcuffed by one pollster artist doing a huge amount of polling and other artists backing away and not doing as much.

    8. So will you place the same caveat on all medical research?

      Exit polls are the basis of the historical information relating overall results to demographic breakdowns. It is an approximation, but an approximation that's good enough to connect one set of data to the other with over a 96% success rate.

  10. Latest Forum poll March 31st.

    Libs 3 point lead and Harper third as best PM.

    1. I don't think anyone believes Harper is well-liked.

      But that poll shows the Liberals down 2 since the last Forum poll, and the CPC down only one. That's a narrowing gap, an improving CPC position overall, and not something that should worry them.

    2. Anything that doesn't show majority worries Harper. And he is no where near a majority.

    3. Wishfful Liberal thinking! It is far from certaain the NDP and Grits will be able to come to a governing agreement. Moew fundamewntally more or both opposition parties may feel their interests are better represented in another election or denying the other the opportunity to govern. It is nothing short of political suicide for the Liberal party to invite the NDP into Government. to do so would split the centre-left-left vote for a generation or more and deny the Liberal party the ability to win a majority.

      In any case Harper is only 27 seats away from a majority with the average and far surpasses irt with the high and max range. The Liberals on the other hand high range is below the Tories' average seat projection and their max only gets them a minority government- Liberals need to start asking themselves if it is time to cease the futility and close shop!.

    4. Also the Liberal's Max seat range, 158, is below the Tories high range, 162 and well below the Conservative max range of 193. The conclusion one must draw is that it is, at this juncture in any case, very unlikely the Liberals can form a Government on their own and that is before one analyses the constitutional hurdles that greatly favour the incumbent and or largest party in the House.

    5. bede,

      You forget the categorical weighting on this site. There is a 58% chance the CPC will finish 143 to 162, a 12% chance the CPC will finish 162 to 193, and a 2.5% chance to finish higher. Whereas the LPC are more liable to finish in the low to average range or high to maximum range than the average to high range.

      Unless there are some fundamental shifts between now and the election, we'll probably see another Harper minority. How he recovers from that will be telling of his future and the future of the CPC.

    6. Being in a majority position now wouldn't benefit Harper at all (because he foolishly passed the fixed election date law). His objective is to be in a majority position on election day.

    7. Mapleson,

      Point taken but, I don't see the probabilities you refer to, where are they?. However, I think if you use qualitative data you'll find that the winds are clearly moving in Harper's favour. For example, today; all provincial governments are majority governments, Harper continually ranks higher than Trudeau on preferred PM questions, or the economy is becoming the number one issue, or the Conservative surge in Quebec in recent polls-all bode well for Harper. Fundamentally, people like Trudeau but, are leery about giving him the keys to the store.

    8. http://www.threehundredeight.com/p/canada.html


      There is a difference between being in Harper's favour and moving in his favour. If anything, I would say we are currently in the middle of a political dry spell.

      1) The fact that all provincial governments are majority governments has no direct bearing on federal elections. Since 1950, we've had 8 minority federal governments ('57 CPC, '62 CPC, '63 LPC, '72 LPC, '79 CPC, '04 LPC, '06 CPC, '08 CPC). Years free of provincial minorities are 1954-1957, 1960-1968, 1982-1984, 1991-1997, 2010, and 2015. Three of these 3 were under the same conditions.

      2) Harper continually ranks higher than Trudeau on preferred PM questions, and Harper continually ranks higher on the disapproval question. The relative worth of these questions is debatable.

      3) I wouldn't say the economy is becoming the number one issue, but has remained the number one issue. We are still waiting on Joe Oliver's first budget on April 21. Doubling the TFSA will pay well with the base, but with only 1 in 3 Canadians having a TFSA, will it be protrayed like the Family Tax Credits using taxpayer money to underwrite the rich? Specifically in the PBO February report “TFSA contribution room limits will soon exceed the financial asset base for most low-wealth to middle-wealth households. Continued growth in individual limits offers no additional benefit for these groups."

      4) The Conservative surge in Quebec in recent polls is significant in that the CPC is now ahead of the BQ, but how much higher can the CPC be expected to rise? Brian Mulroney was the last conservative to take a majority of the province. Since the rise of the modern CPC, the results have been 8.8%/0 seats in 2004, 24.6%/10 seats in 2006, 21.7%/10 seats in 2008, 16.5%/5 seats in 2011. So the "surge" is really a return to "normal" levels.

      This month will be a telling one with whatever negative press there is with Duffy and the favour of the budget.

    9. Mapleson,

      You have exaggerated the Liberals potential: The Liberal party is more likely to finish in the low-average range but, only has a 25% chance of finishing in the high-max range.

      I'm sorry you can't see what is happening Mapleson, Harper is well on his way to a second consequtive majority and Trudeau with a 50% chance of underperforming the seat average well on his way to destroying the Liberal party and becoming the third party leader once again

    10. While Trudeau may not be the dynamic leader needed to win a majority, I am at a loss as to seeing Harper retaining a majority. I feel a minority government will be in place, party undetermined at this time.

    11. bede,
      How did I exaggerate? I specifically said "are more liable to finish in the low to average range or high to maximum range than the average to high range."

      Currently, there is somewhere between a 1 in 8 and 1 in 10 chance of another Harper majority, by the numbers. Any interpretation beyond the numbers is opinion. I haven't said anything here that isn't directly related to the numbers or indicated as an inference.

    12. Mapleson,

      The Liberals are not more likely to finish in the high-max range. They only have a 25% chance of doing so! To say they they are more liable (sic), I strongly suggest you look up the meaning of "liable" because it is a poorly chosen word in the context you use it in, I would say incorrect, hence my (sic). Is incorrect. They are only possess a greater probability to finish in the high max range in relation to the Tories or Dippers but, even this is open to interpretation because they are grouped in with the minor parties! A fact you have conveniently excluded! In fact we don't know any of the Liberal's probabilities for any of the ranges because of this grouping and that makes your statement subject to interpretation and analysis in relation to these minor parties, in conjunction with more research to separate the Liberals from them.

      Dave MacKay,

      The last poll I saw had the Tories up four points on the Liberals in Ontario. Ontarians are not rally en masse toward Trudeau and the Liberal party in fact they seem to have great hesitation toward young Trudeau. The NDP is on the rise, such a rise in popularity disproportionately hurts the Liberals and in doing so helps the Tories. People don't love harper but, the don't trust Trudeau and they have a dislike of El Beardo, in uncertain economic times they'll pick stability over youthful inexperience and spendthrift socialism. The other point I bring up that mapleson has down-played is Quebec. In fact the Tories have polled higher than the "normal levels" Mapleson has indicated and as Mulroney commented to Lord Black the other week "all French-Canadians are nationalists" and I think people forget just how discredited the Liberal party and the Trudeau name are in Quebec. Pierre trudeau was suppose to being Quebec into a modern confederation, instead his own hubris and ego destroyed the Constitution and Quebec was left not simply outside the document but, betrayed since, their historic veto was expunged by the Courts. Ontarians were willing to vote for the Liberals so long as they could win Quebec and keep the country together-that no longer looks to be the case and therefore it is no coincidence the last poll from Forum had the Tories with a 4 point lead over the Liberals in Ontario.

    13. bede,

      25% is more than 20%. Liable, beyond the legal definition, means likely to do or to be something. I have obscured no facts, only pointed out the construct of the model. The Liberals are a third party at the moment, as the NDP were in the last election cycle. If you have a problem with the Eric's model, take it up with him. I am only using it for the basis of discussions on this site. Eric feels that the model is sufficiently complex without having a fourth category splitting "other major parties" and "other minor parties" (please correct me if, I misspoke, Eric).

    14. bede,

      Why cherrypick polls that are to your liking? The 308 model has the CPC up by 1.4% in Ontario, and taking the plurality of seats. Taking a majority of seats is necessary for another majority. This would require both more equal competition between the LPC and NDP and the CPC pulling in more LPC/NDP voters. The question is if there are more people who dislike Harper or more people who dislike Trudeau.

      You can imagine a popular uprising in Quebec to support Harper all you like, but there has been no indication that Quebec has changed from the pattern of the last 6 elections. The fact is that they are running third and only slightly ahead of the BQ in Quebec.

      Ontarians are more concerned with Ontario than the majority of seats in Quebec. Otherwise, why isn't it reported on the local news?

      You present no facts, only rhetoric.

    15. Why cherrypick stats or fact Mapleson? The Grits are grouped with the minor parties-you have obscured facts not observed because you present these facts outside their proper context to the benefit of your own particular point of view. The Liberals do not have a 25% probability to finish in the high-Max range; the Green Party, BQ, and Liberals have a 25% probability of doing so. That is entirely different than what you wrote and it makes your original writing disingenuous. That I trust was an unintentional outcome.

      Liable has the meaning of; "likely to do something" but, usually when used as an adjective in such form a description of the activity follows. In any case the first definition is the legal term so, you are really stretching and the way you use liable more closely resembles the third definition when followed by "to"; for example: Liable to flooding.

    16. I would have mentioned the Greens, BQ, or NDP, if they had been pertainent to the discussion. I apoligize if anyone was misleadby my assumption that people here had read how the 308 model works.

      My original phrase was "more liable to" so I would say it exactly resembles the third defintion, by your ordering. It was used in the context of A has a higher probability than B. From all the push back on my wording, rather than my logic, it was obviously a bad choice.

  11. Anything to reject the unfortunate results I see.

  12. Wow, looks like Nanos has CPC dropping quite a bit in all categories. I guess scaring Canadians is not working so well.

    1. I don't see a Nanos poll asking those questions.

      I do see one pointing out that terrorism is a niche issue, but I think everyone already knew that. The CPC focusses on it because they think it's an opportunity for gains for them, with limited opportunities for losses.

      Much the same way that all major political parties in the US oppose gun control, even though a majority of Americans support it. They gain more net votes by opposing it than they would lose by opposing it.

      I expect the CPC has done similar math pursuant to terrorism.

    2. Oh, there it is. Now I see it.

      Harper still the preferred PM, though, according to Nanos.

    3. Not by much and he is falling. Forum has him third. I suspect it is somewhere in the middle and falling.

    4. Don't put much stalk in a rise or fall of a point or two on any poll to poll change...unless its outside the margin of error then it is at least worth considering it to be actual movement rather than statistical noise. As always it is the long term trend that is going to be far more valuable rather than a single month here or there.

    5. Yup, long term trend shows no majority for Harper. His tenure is coming to a close.

    6. "Harper still the preferred PM, though, according to Nanos."

      By less than one percent LOL.

      According to Nanos, Harper and Trudeau are in a virtual tie. This is what he says.

      "After a series of weeks of positive momentum on the Nanos Party Power Index, the Conservatives
      registered a decline and still trail the Liberals. Also of note, the preferred Prime Minister numbers for
      Harper also declined. Tories declines cut across all demographic groups."

    7. @Carl Szczerski - Like that steady decline the Liberals have been in since October?

  13. This comment has been removed by the author.

  14. Eric,

    The Calgary Herald reports the writ will be dropped tomorrow for an Alberta general election.


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.