- Date published:
9:30 am, October 14th, 2017 - 160 comments
Categories: Deep stuff, election 2017, winston peters - Tags: coalition talks, election 2017, exercises in futility, psyc101, social science, winston peters
I am cursed with the wretched combination of holding impassioned Red/Green values in tense interaction with my decades of work as a social scientist. Every morning, as the waiting for Winston to decide continues to fray my fragile psyche, I can look at myself in the mirror, contemplate my own frenzied speculative hopes for an Ardern Prime Ministership, and immediately think of a host of strong social-methodological reasons why all such speculation is futile.
There are a host of reasons, but I’ll concentrate on three:
1) Categorical over-reach (ie. There is ‘one true meaning’ to what has happened in this election).
Elections are won and lost over a massive array of complex factors. Their outcome is co-produced by endless different, complex interactions. Just like sporting events, we can only comprehensively place a narrative framework on them once the game is over (and even then, by selecting out particular elements of what has happened). Claire Robinson in The Spinoff, fell into exactly this error by declaring this to be a ‘status quo’ election based on one factor – the state of opinion polls 18 months prior to the election – and then retro-fitting this as signifying ‘no change’ as the one true meaning of the election. Andrew Geddis did a good job of rebutting this search for single categorical understandings of what is happening.
But we are all prone to this tendency: picking one single dynamic and using it as the meta-cause beyond which all other things are merely effects. I’ll get back to this problem in a moment, but first, the challenge gets further complicated by another problem…
2) Confirmation Bias (I really don’t need to explain this because we know what this is…):
Bryce Edwards’ columns in the Herald listing ten reasons why the result will go National’s way and then 13 reasons for a Labour win, could be interpreted in a number of ways. As a person who strongly desires one over the other of these outcomes, I read these in very different ways. When I was reading the case for National, I wanted to see the counter-argument to every one of his ten pieces of evidence. When I read the case for Labour, not only was I convinced well before I read the whole thirteen reasons, I didn’t feel the need to see countermanding evidence for even one of them! Thirteen versus Ten reasons! Count em!
Nowhere is this confirmation bias more evident than in our parsing of the character and motivations of Winston Peters himself. US progressives have recently been through a rollercoaster ride of love/hate with John McCain. In the lead-up to his fateful vote ‘saving’ Obamacare, he filled the classic position of the centrist ‘maverick’ politician: there was plenty to hate and nevertheless a lot to admire and like. The moment his vote went in our favour (I’m assuming all Standardista’s wanted to retain Obamacare), his actions and motivations were re-cast in a stellar new light.
Peters gives us an even more complex tableau because his capacity for making multiple contradictory statements, and the sheer volume of his pronouncements have created a whirling vortex of possible confirmation bias for those of us trying to read his mind. There will be an outpouring of love/hate for Winston after this decision with little in between. Maybe it will be best for the emotional stability of progressives if we prepare for either option?
Which leads to the addition of a third layer to the ‘impossible to guess’ cake that I am trying to bake…
3) Elections are Complex not Complicated.
This is one of the key distinctions across a whole pile of methodological fields. Complicated things have multiple independent moving parts which than combine to create a specific effect. Space Shuttles are complicated. There are millions of moving parts, but if you have the blueprints, you can reproduce one and get, more or less, the same thing. Complex systems, however, are defined by sets of relationships that are not independent to each other. They move and change – which then changes other outcomes across the entire system. The relationships between parents and children are complex – they involve multiple relationships that most definitely change over time and produce unpredictably complex effects. There is no blueprint for a successful parent-child relationship.
This critical methodological distinction is the reason why Nate Silver’s 538 model – which was designed to predict the US election outcome in 2016 – gave Trump a 30% chance of winning (actually, quite a distinct chance) compared to the Princeton Electoral Consortium model’s prediction which gave Trump no chance at around 1% (which, BTW, confirmation bias then amplified to be the ‘correct’ prediction for progressives evaluating these two models). The Princeton model was complicated – it evaluated state by state polling with each state being independent from the others. Margins of errors therefore cancelled themselves out over all the states (ie. If a poll had a margin of error that put Trump up too far in Minnesota, random error would balance that with the opposite margin of error in another state). Nate Silver’s model predicted that bunches of states move in complex connection to each other – so if a polling shift in one state showed a bit of potential movement towards Trump, he extrapolated that to all the connected states in his model, amplifying the potential shift. Hence a shift in polling in one Rust-belt state indicated a shift in them all. As we all know to our great regret – the complex model was far more accurate on the night. Trump didn’t win at 100-1 odds, he finessed a one-third chance at victory via a complex shift in voting behaviour across the Rust-belt.
So, elections are not complicated they are complex. If we run down Bryce Edward’s lists of factors supporting an outcome in favour of National or Labour – they aren’t causally-independent from each other. Movement in one factor will potentially influence sets of others. They are in complex interaction. We don’t know which the ‘prime moving’ factor will be that will trigger decisive movement across all the others.
That won’t stop us trying, but really we can’t know which is the primary motivator in Winston’s mind and we can’t know how that will interact with the wider caucus and, now, the NZ First Boards as well. Finally, and most important of all, the prime factor may be something that has been offered or said inside the confidential meetings which can have no way of knowing until well after the fact.
Unfortunately, for the truly obsessed, you can’t simply eliminate confirmation bias by going down the list of ten/thirteen factors that Edwards identifies and add up which are less likely than others and give an accumulated likelihood score of one overall outcome over the other. Maybe we should just add them all up? Maybe thirteen is clearly better than ten? As we know from studies of jury behaviour, for most jurors (the ones who actually end up making a decision one way or the other), the evidence accumulates in complex patterns and ONE piece of evidence ends up being the tipping point that shifts the outcome one way or the other. And for many jurors in the same cases, it is different pieces of evidence that tip them over. We just don’t know what that one tipping point factor will be for Winston. Not that we’ll stop trying to guess!
4) What we do know: it is a tight decision.
In all this murkiness, one thing has become more clear over the last few days. There is one fact we can hang some hats on – Winston has missed his deadline!
If there was one clear factor driving the subsequent complexities in one direction, or if Winston had entered this process with a clear destination in mind, the game would be over by now. He had ample time and information to justify a decision and make his announcement yesterday.
The fact that he has missed that deadline comes down, in my mind, to one of three things:
Of these, only the second scenario would suggest that this is a clear-cut decision. I’d be very interested in any other suggestions Standardista’s might have, but the half of me governed by social-scientific sensibilities is winning over my insane Red/Green optimist… This is too close to call and we cannot in any way guess the outcome.