There is a saying attributed to former Canadian Prime Minister John Diefenbaker. When asked about polls, he employed the homonym of the word and declared that they were for dogs as they were the only ones who knew how to properly use them.
Polls do get a bad name – sometimes deservedly so, and many times not. Sometimes they capture the essence of a race and pick up the trend, and other times they are wildly off. In any given race, you will get different polls that say very different things.
Whether it is the prospects of Boris Johnson’s Premiership, or the current election here in Canada, people follow the polls and have very animated opinions about what they mean, or not.
I have had some limited and passing experience with polling, so I am sympathetic to the companies and do feel that, by and large, they endeavor to do solid work. And why wouldn’t they? Bad data makes you look unprofessional and incompetent – hardly a winning business model.
Every player in the industry wants to do their best work, and they go through a great deal of methodological review to be as accurate as possible. The problem is that they need to rely on subjects that can be as changeable as the weather.
A poll is a snapshot of a body in motion. Like a still photo of a race at the mid-point, it gives you the relative position of each ‘runner’, inferring the action up to that point, and mildly predictive of what comes next. It can’t show you that within ten yards a runner may stumble, or another gets their second wind. All it can do is tell you that on a given moment in time, that is where they all are. Blink, and it can change.
It could be that our expectation of wanting to know the future makes us project more meaning on individual polls than we should.
Polls have a function, and a value, but not in predicting outcomes. That is more for those who like the excitement of a race, using them as a short-hand to keep score. For me, they are a means of testing hypotheses – of figuring out whether or not one’s gut intuition is reliable. On that score, the polls are doing their job – and I respectfully disagree with them.
Right now, they show a tight horserace between the Tories and Liberals, with the alternating leads falling within the margin of error. When extrapolated into seat counts, the Liberals get on average 10-15 more, owing chiefly to what the political scientists call a ‘more efficient vote’. It is code for the Liberals getting growth in competitive ‘ridings’ (Canuck slang for constituencies) while Tory increases come in areas that they were going to win big in anyway.
While there is still almost a month left, and anything can happen, I am going to boldly (and maybe foolishly) predict that the polls are off and that we are looking at a Tory minority (‘hung’) parliament, with a seat count in the range of 163-168. That is at wide variance with what is currently being estimated, but I think it is possible. If it happens, it would be a seat-for-seat reversal for the Liberals and Conservatives based on what is being estimated.
This is why I think this could happen.
When building a model for a voting population, one has to rely on statistical data on hand – measurable and verifiable. That means building a sample that reflects the demographics of the country. In rough terms, if men aged 18-34 are 25% of eligible voting population, they have to be 25% of your sample, and so on. That doesn’t mean they are actual voters – only that they are legally permitted to show up on polling day and mark an X on a ballot. They are ‘potential’ voters.
In 2011, Stephen Harper won a majority government, and in 2015 lost it to Mr. Trudeau, who grabbed enough to secure a majority of his own. Conventional wisdom says that the Conservative vote collapsed, and yet the difference in total national popular vote for the Tories dropped by about 219,000 votes – a 7.7 percent decline that cost the party 60 seats. By contrast, the NDP lost fewer seats (51) despite having a drop in popular vote that was well over 1 million.
The Trudeau Liberals vaulted from 3rd place to majority status on the strength of an increase in their vote in the range of 4.2 million more voters than in 2011. Of course, if you’re doing the math, you know that if the Liberals took every lost Tory and NDP vote, that still leaves a gap of a little over 2.9 million extra votes unaccounted for.
So, where did all those people come from?
The number to bear in mind is 17.4. This number was key to making Mr. Trudeau Prime Minister, and any decrease to that number changes the math dramatically. It’s a percentage and represents the change in voter turnout from 2011 to 2015, but not for the entire electorate. It applies specifically to one demographic – the 18-34-year olds – millennials. (Graph below)
Those aged 18 to 34 turned out in numbers that were wholly out of keeping with past elections. While many other demographics were up, such as over 65’s who were up about 7%, it was the millennial vote that took off like a rocket. As younger voters generally skew Liberal, this increase had a powerful impact on both the popular vote and the seat count. In short, he was exceedingly effective in converting potential voters into actual voters.
But it’s a double-edged sword.
The 18 to 34-year old demographic views politics differently, but not in the way you think. We get so used to pin pointing stances on individual issues that we commit the error of ‘not seeing the forest for the trees.’
Think in terms of ‘abstract’ versus ‘practical’. The abstract group views issues largely in terms of theoretical constructs. That is, they are more ideological. While not partisan, they do subscribe to labels like ‘liberal’ and ‘conservative’ in a more holistic fashion. It’s a way of life, a lens through which one views the world.
Practical voters are not necessarily partisan either, but they view the world in terms of immediate concerns. There may be a ‘unified general theory of society’ running in the background, but if their political GPS is going to direct them into a farmer’s field, they will drive on instinct and ignore the gadget on their dashboard.
Young voters are abstract, while older voters are practical. One worries about whether they will be able to survive twenty years from now, while the other worries whether they will be able to survive until their next pay day.
Abstract voters are naturally driven by big ideas, and big personalities. Justin Trudeau is, admittedly, the perfect avatar for abstract voters. It is less about policies and more about lifestyle, a kind of culture. It is about the ‘brand’. This is not to suggest that abstract voters are not serious people – they are – but they are more about the ‘macro-ideology.’
Practical voters have some affinity for this, but they are more concerned with the fine print and the price. They’ve spent years dealing with how to make pay cheques cover housing, transportation, food and raising children. Life has made them cynical, as it does all of us at some point. They are ‘micro-ideological’. They will join on the journey to the New Jerusalem so long as there is a job and a place to live when they get there.
The polls reflect two things – ‘eligible’ voters (share of the general population), and ‘likely’ voters (those who are predicted to actually show up). Census data can give you the first number, but for the second one, you have to rely on the past, and that includes the last election where one demographic saw a huge spike.
If you look at historical data, for the 18-34 age group, you find that since 1965 there were only two times that their turnout was high – the last election under Justin Trudeau (first graph above), and the period between 1972 and 1980, under his father. The 1984 election which brought Brian Mulroney and the Progressive Conservatives in with the largest majority in Canadian history saw the participation rate for that group drop off a cliff, while the group of voters aged 55 and older saw a respectable increase (Graph below).[i]
Trudeau Senior kept his brand intact, but it was the consequence of economic decline, massive deficits and a decade of power (notwithstanding the Joe Clark interregnum) that tipped the scale. Trudeau Junior does not have a dramatically bad economy, but it is bad in key areas – and he has the deficits.
The real question is whether or not he still has the brand.
If he does, and the abstract ‘meta-ideology / world view’ holds, then 18 to 34-year olds will come out in similar numbers. But the Trudeau brand has to contend with at least three instances of brownface / blackface, the SNC Lavalin controversy, the treatment of Vice-Admiral Mark Norman, as well as former Ministers Jody Wilson-Raybould and Jane Philpott. And the 18 to 34-year old voting demographic is the most volatile, being the most reactionary to negative campaigning and ‘inauthentic’ behaviour.
In short, as quickly as that group turned out for him in 2015, they can also turn away just as fast. And if that demographic is disillusioned enough to shift to the NDP or Greens – or just stay home – seat counts shift wildly.
This says nothing of the older demographic that may feel that the combination of virtue-signaling, lack of probity and declining economic prospects incentivizes coming out in good numbers.
Do not think that the parties do not understand this. Look at how each of them has framed their narratives and you can tell that they are trying to firm up their ground. Trudeau is campaigning on the macro-message (the awkwardly constructed “Choose Forward”) while the Tories are playing the micro-message (“It’s time for you to get ahead”). The NDP may very well have the perfect message, even if their policies are not great – “In it for you”. It is the perfect marriage of the macro and micro. Big picture enough for the idealistic, but immediate enough to answer individual concerns.
It won’t be enough to make Jagmeet Singh Prime Minister, but it’s enough to give Justin Trudeau issues. This says nothing of the far more substantive reputation of the Green Party when it comes to the climate change issue. The Prime Minister can try to ‘out green’ the Greens, but it is a strategy that would have very limited success.
In terms of the main contest for government, when the Liberals talk of cuts by provincial Conservative governments, they are trying to undermine the Tory message with ‘practical’ voters. When the Tories talk about scandals and instances of implied hypocrisy, it’s aimed squarely at shaking the ‘abstracts’.
The best that the Liberals can hope for is some peripheral pickup of Conservative support or convince older voters to stay home. For the Tories, the options are greater, as they can simply nudge Liberal supporters to either the NDP or Greens, in addition to sitting this one out. A more volatile demographic with a low tolerance for negative campaigning and hypocritical actions, and with more options and alternatives to voting for the Liberals – this is the dynamic that doesn’t get captured fully in the polls – or at least the ones you can read for free on a news site.
And so, my unscientific gut feeling is that the people who want to punish Trudeau have more incentive – and alternatives – than those who want to keep him. I believe that millennial voter turnout will drop to more traditional levels and that older voters will come out at the rate they’ve been coming for years.
A quarter of the Canadian House of Commons will likely be decided on marginal votes, so if this scenario happens, the polls could be very wrong. Of course, 2019 could be a repeat of 2015 and this whole article would have been a waste of your time and mine.
October 22nd will reveal which is true.