Blame the Polls for Trump

Here we go again. The polls unanimously predicted a victory for Clinton, and we all know what happened. The polls also unanimously predicted that Britain would remain in the EU, but they were wrong there too. Only months earlier the UK polls predicted every sort of party combination to form a government, except for the one that won.

Why do polls get it completely wrong, especially when they're unanimously predicting a close result? Perhaps it's because of the Observer Effect (sometimes confused with the Heisenberg Uncertainty Principle). Simply put, it states that by measuring something, you can affect it. In the case of opinion polls, this is what happens:

1. The polls all predict a close result
   
2. If the person or party you like is losing, you become more likely to vote because you can make difference
   
3. If the person or party you like is winning, you become less likely to vote because you don't need to

There's a sweet spot for this effect. If the polls are predicting the person or party you like is a long way behind in the polls, you might doubt your own judgement and you might not want to waste your vote. The further behind they become in the polls, the less likely you will be to vote. Equally, the further ahead the person or party you want to win becomes, the less likely you are to vote because your vote is probably not needed. So the effect of a poll that's not close is to reduce voting at both ends of the scale (as we see in typically low turnout UK and US elections). The results of such polls are consequently more accurate than when they are closer. The polls have had an 'observer effect', but it's equally applied to both sides so not noticed in the result.

If, on the other hand, it's very close so much so that the polls themselves become divided, then both sides will be encouraged to vote - probably in equal proportions. The result will then also be close to what the polls predict. The Observer Effect has increased the volume of voters, but the vote will remain tight.

So the Observer Effect on polls is only evident where the polls unanimously predict a reasonably close win for one side. Supporters for the side predicted to lose will believe their votes will make more of a difference than the supporters for the side predicted to win. This sweet spot is probably around 5% margin between the candidates. Which means it needs only a 2.5% increase in predicted loser voter numbers to reverse the prediction. It wasn't god supporting Trump (as I heard some whackos claim this morning), it was maths and the polls.

If I'm right and this is the main reason we have a bozo as the leader of the free world, then instead of agonising over what policies and hairstyles our candidates should adopt in future, we simply need to stop doing polls! Or at least agree amongst all the mainstream media to ban them for a couple of weeks before the vote. Don't forget that if you think America looks stupid this morning, the media and 'experts' look even more incompetent. It's in their own interests to prevent this from happening again.

No doubt when my blog post goes viral, I will be vilified for #BanThePoles, but that too is bound to be on the orange chap's agenda at some point.