Making emotional predictions

“Forecasts are always wrong,” said Robert Chote, chairman of the Office for Budget Responsibility, this week (having said similar things in the past). This is a brave line to take when your organisation produces the economic forecasts used by the government, but it is also true.

In anything that involves human behaviour, the best a forecaster can do is assess the situation in the recent past and at present, note the rates of change, then take a view on whether things will continue to change at those rates or if there are good reasons otherwise. It makes sense to say how confident you are in the prediction or provide a range of possibilities.

This model can work reasonably well in the economy, at least in the short term, but has had a very bad year on predicting the outcomes of votes. The Register has just published my article on Loughborough University’s work to track the public mood through emotions in tweets. It has found that stable levels of emotion are good for a candidate or a party, and its model put Donald Trump ahead on emotional stability (in tweets about him, anyway) for almost all of the three weeks before polling day.

This year it has seemed as if people are taking massive decisions based on emotions alone. I’m not convinced – votes have always been made on gut reactions as well as cold analysis. But if millions are willing to use social media to broadcast their emotions, which are obviously a factor in most decisions, it makes sense to track them.

This is from my monthly newsletter, which you can see in full here. Sign up below.


This entry was posted in Articles, IT, Media, Politics. Bookmark the permalink.