18 November 2014

The (only) Question that Could Have Accurately Predicted the Winner in Recent Elections in Romania

I guess (some of) you know that, last Sunday, Romanians elected a new president. This is not news.

The news is that the under-dog challenger won against all odds and predictions.

All pre-election polls predicted that the favourite – the incumbent Prime Minister – will win and at a comfortable difference.

But, things didn’t happen as expected. The opposition’s candidate won at a very comfortable difference – 54.5% to 45.5% (roughly 9% of expressed votes) which in absolute numbers represents more than 1.100.000 votes.

Now, that the results are known, there are a lot of people explaining how it happened.

Yeah… Hindsight Bias!

There is, however, a question – methodology – that could have predicted the winner with accuracy.

Which of The Two Candidates Looks More Like a President?

Which of The Two Candidates Looks More Competent?

If you answered the one on the right-hand side, you were correct. He was the under-dog, but still won with 54.5% of votes.

The two questions above seem trivial and even irrelevant, but this is not exactly true. They are based on the Representativeness Heuristic

Basically, instead of answering the difficult question of 

“Is this guy competent” 

We answer the simpler one of 

“Does this guy look like a competent person”.

Todorov et. al. (2005) found that asking the questions of “who looks more competent?” gives accurate predictions on who will win the elections in 65% of cases.

A. Todorov, A. N. Mandisodza, A. Goren, C. C. Hall, (2005) Inferences of Competence from Faces Predict Election Outcomes, Science 308, 1623. 

No comments: