The faces are different, but the gestures are the same. A palm clasped over the mouth, fingers dragged through hair, the back of a hand wiping a runny nose. The images of Hillary Clinton supporters of what was supposed to be her victory party in 2016 are a study in shock. Polls had given Clinton a 70% chance of victory, some even more.

The result made pollsters look foolish.

Pundits began to speculate that Trump-era politics defied prediction.

But in 2018 academics at the University of Southampton and the University of Texas tested the

accuracy of opinion polling around the world since the Second World War. It's been remarkably consistent.

Hillary Clinton did win the popular vote after all. With 143 days to go this is

checks and balance. and This is a podcast about the 2020 elections.

Each week we take one big theme shaping American politics and explore it in depth.

Today. How useful are the polls?

This week the Economist launches its first ever statistical forecast of a US presidential election.

In this episode we'll speak to the economists data team to find out how to build an election

predictor, how confident they can be that it will work and who's currently predicted to win the White House in November.

We'll also hear the story of George Gallup, the pioneer of political polling, and speak to a former pollster for Barak Obama about how to use

polls when public opinion appears to be so volatile. With me as ever to discuss all of this are Charlotte Howard, the economist's New York

Bureau Chief and John Fasman, the Washington correspondent.

How are you both doing?

Doing well, John, thank you.

In New York, protests continue mainly peacefully.

I have someone I went to high school with who was arrested though

while protesting perfectly peacefully and attacked by a police officer on Saturday night. So it was just a reminder of how violence continues

often out of the public eye and how important it is for these protests to continue.

I'm sorry to hear that.

...