Wednesday 9 November 2016

Here is what just happened

I am writing this on the morning after the US election so I don't think I need to explain the headline....

I am sure there will be more sophisticated explanations emerge over the following day but this is my take to understand the result of the election.

In the run up to the US election out team have been conducting an ongoing series of political polling experiments to understand what has been going on. We have not been polling in a conventional sense, more experimenting with how to measure voting sentiments and voting intention in an attempt to find a better way of predicting the outcome of tight elections.

We have been asking a lot more indirect question about the difficulties people were having in making up their mind, exploring more implicit measures of voter sentiment to see how they stack up with declared measures, getting people to play games to predict the outcome to reveal some of their hidden feelings and we have also focused a lot of attention on asking open ended questions to measure the level of passion in the arguments and look at what reasons people have been using to make their choices. 

Here is what I think has just happened from my perspective from the learnings from all this research.

Trump messaging was far stronger than Hillarys from the get go. It was more coherent. Make American great again. Close the borders. Drain the swamp, Hillary Clinton is corrupt.  We were seeing all this being echoed back over and over again to us in the explanations gave us for why people wanted to vote for him. 

On the other hand Clintons messaging was, extremely weak, in fact almost none of it seemed to stick. Less than 10% I estimate of the reasons cited for voting for Clinton were anything to do with liking her policies, it was nearly all to do with stopping Trump winning. So many people caveated their choice with an explanation that they were picking the lesser of 2 evils.

All the implicit candidate favourability measures we undertook showed us how much Clinton grated on the American public. Flashing pictures of her face solicited up to a -60% negative reaction, even worse that Trump who is known as being a pantomime villain.  Here face and perhaps more importantly her voice did not fit. The majority of American public implicitly did not warm to her. 

What Trump was up against in this campaign was not Clinton, but Trump himself and, well let's not beat about the bush here, his personal sociopathic character traits.

The Trump sexual harassment scandal embodied all this and the misgivings so many people were having about handing him power seriously pegged back Trumps latent momentum in the final month of the campaign as all this news broke out. 

But beneath all this, he was a lot of people's implicit preference,  but they could not express that in opinion polls or even to themselves to  that matter due to the outrage being voiced in the media about his behaviour, but his core message resonated.

Then comes along the FBI email investigation a week or so before the election. This was like pulling out literally a “Trump” card.  What is did was give all the people who latently liked his messaging but were suffering from cognitive dissonance over his character, a strong emotional counter argument for prefering him it gave them something that they could dress up as a lot more significant, it validated all his messaging about his opponent too. It could not have been more perfectly timed.

We  actually saw the change happening in real time as we conducted a large scale research experiment the week before the FBI press release so on the Monday after this we did a follow up piece of research to see what was happening and  Clinton's 7 point lead we had seen the week before literally evaporated overnight. 

The shy Trump supporters were released from the closet so to speak and at the same time all the people who didn't like Trump or Clinton were given a strong reason not to vote for either candidate or stay at home and not vote at all. 

The last minute FBI volte-face was far too late in the day to undo any of this.

What we have learnt from all this that the public's opinion on how they will vote is very emotional process,  similar to the way my daughter makes up her mind about what type of curry she wants to order in our local curry house. She tries on several  choices to see how she feel about them thinks she has made up her mind, changes it, changes it again and then ditches them at the last minute when the waiter is standing over her and decides with 
her gut instinct.

In this case the US gut fancied trying something new, despite some serious misgivings  because the other dish did not seem too appetizing at all they decided to roll the dice.

You can chide the pollsters if you like, but this type of emotionally charged election is almost impossible to predict even a day or so out but certainly its clear you cannot predict an election by simply asking people which way they intend to vote.  


Here are a few other closing thought about why the polls got it wrong....

1. Are the types of people that slam down the phone if a pollster calls and would never think about doing an online survey also the same type of people who might be more likely to have voted for Trump?

3. Clearly Hillary Clinton did not motivate here democratic base to vote in the same way Trump rallied his supporters and so the polls which are often weighted by past voting activity were delivering a miss read

2. Likewise did weighting polls by traditional political allegiances have any relevance in this election

4. Male blue collar workers who voted for Trump in droves are the hardest group to reach with research as they are working

5. There is some evidence in our research that Trump supporters were slower to respond to our online poll invitations and so some short turn around polls might have closed up before all the Trump supports had a chance to register their opinion.


No comments:

Post a Comment