##
How do you predict an election?

How to predict an election. Most of us have a vague idea of how you might go about predicting the result of this week’s general election: go out and ask a large and random collection of people who they intend to vote for. The share of the vote each party gets in your sample of people reflects the share of the vote you can expect nationally.

##
What makes election forecasting so different from other applications?

What makes election forecasting very different from many other other applications is that we’ll soon find out, unequivocally, whether its predictions are correct: this will probably be some time in the early hours of May 8th.

##
How do you calculate the probability of winning a general election?

The aim is to produce a probability of each candidate winning in each constituency. To work out the probability of a party winning the overall election, you add up the probabilities of it winning the individual seats.

##
Where do election forecasters get their data from?

The forecaster ElectionForecastUK, for example, uses data from YouGov and all publicly released national and constituency polls. They combine this with historical election results going back to 1979, as well as data from the UK Census. The general approach (nicely explained on their website) is this.

Why is the outcome of the election different each time?

The outcome of the election will be different each time** because the virtual people’s behaviour is governed by probabilities, not certainties. **

What is the probability that a randomly chosen person will vote Labour?

If out of 100 people 25 say they will vote Labour, then your estimate of the probability that a randomly chosen person will vote Labour is** 0.25 **.

What is Bayesian analysis?

Bayesian analysis is** being used in all sorts of areas, not just election forecasting. ** Spiegelhalter has developed a computer language called BUGS, designed to be used in the biomedical sciences. An alternative incarnation of BUGS, called JAGS, is being used in election forecasting.

Is the election decided on the national share of the vote?

The** election isn’t decided ** on the national share of the vote, but on the number of seats, that is, on who wins the majority of votes in each constituency. So you’d really have to go and poll a large number of people in each constituency — and that’s a lot of polls to conduct. This is why election forecasting is a very sophisticated business.

Why are Maine and Nebraska shown separately?

Maine and Nebraska’s congressional districts are shown separately** because those states split their Electoral College votes, allotting some to the statewide winner and some to the winner of each district. **

Is there a 10 percent chance of winning?

A 10 percent chance of winning is** not ** a zero percent chance. In fact, that is roughly the same odds that it’s raining in downtown Los Angeles. And it doesrain there. (Downtown L.A. has about 36 rainy days per year, or about a 1-in-10 shot of a rainy day.)

Is there a chance of polling error?

Of course,** there’s always the chance of a polling error, which tends to be correlated from state to statewhen it happens. ** Trump needs a bigger-than-normal errorin his favor, but the real possibility that polls are underestimating Trump’s support is why he still has a path to win reelection.