What is an exit poll? Scientists reveal if the prediction tool is accurate - as voters cast their ballot TODAY

What is an exit poll? Scientists reveal if the prediction tool is accurate - as voters cast their ballot TODAY
By: dailymail Posted On: July 04, 2024 View: 95

As Britons flock to cast their votes in the 2024 General Election, the nation will be holding its breath awaiting the results.

While true political diehards might stay up all night watching the votes roll in, the exit poll at 10pm tonight will give many their first hint at the results.

The exit poll is conducted each year by a consortium of broadcasters as an early prediction of how the vote may have swung.

Unlike a regular poll which asks voters what they might do in the election, the exit poll asks people who they have voted for as they exit their polling station. 

Luckily for those with an early bedtime, scientists generally agree that the exit poll is a fairly accurate representation of the real vote. 

Today voters around the country will be queuing up outside polling stations. But for those who don't want to wait for the results to come in the exit poll will be released at 10:00 PM

The exit poll is jointly funded by the BBC, Sky News, and ITV News with the actual polling being carried out by research firm Ipsos.

This year, Ipsos is sending interviewers to 133 polling stations around the country.

As voters exit the polling station they are asked to fill in a replica ballot exactly as they have just done and place this in the pollster's ballot box.

According to Ipsos, four in every five people approached at the 2019 General Election agreed to complete the mock ballot paper – so some do refuse to divulge.  

Overall, Ipsos says it will ask about 17,000 from all around the country to gauge the swing between different political parties. 

An exit poll asked voters, like Rishi Sunak pictured here casting his vote, how they have just voted. This is different to a regular poll which only asks people's intentions

These results are then collected and sent securely to a secret location in London where statisticians and analysts have been sequestered away from media influence. 

This is believed to generate a more accurate prediction of the election results than pre-election polls. 

Michael Clemence, from Ipsos UK, told Sky News: 'And also we're dealing with people's behaviour, so we're not asking people how they intend to vote.

'We're talking to electors who just voted, and I'm asking exactly what they just did – so you've cut out the error in prediction polling.'

Pre-election polls can be less accurate since even small factors like the order of names on the ballots can bias people's decisions when they're in the booth. 

However, by asking people what they have already done this removes any potential bias the interviewer might include.

Likewise, Dr Hannah Bunting, co-director of The Elections Centre at the University of Exeter, wrote in an article for The Conversation that exit polls get around many of the problems faced by political polling.

Experts say exit polls provide a generally accurate first impression of how the vote may swing. This year pollsters will interview about 17,000 voters, just like Kier Starmer pictured here casting his vote

Dr Bunting writes: 'They’re repeating an action, instead of imagining something they might do. 

'We know they’ve voted instead of having to estimate their likelihood of turning out.'

The data collected from these interviews is then analysed by a team of experts in the capital.

The current exit poll model was devised in 2005 by John Curtice, professor of politics at the University of Strathclyde, and David Firth, professor of statistics at the University of Warwick. 

Professor Curtice has claimed he already knows the result of the exit poll at 2pm on polling day – but if he told anyone he might end up in jail. 

Although millions of people are still to vote at 2pm, it's thought this may be late enough to work out the results from key seats. 

Since exit polls survey what has already happened they are less influenced by bias and uncertainty among voters

Since pollsters only conduct interviews at a handful of constituencies across the country they need to look at trends and patterns to predict how other regions will vote.

Ipsos spreads its questioning across a variety of urban and rural areas to examine voter swing.

Marginal seats are also more closely examined in order to make predictions on key electoral changes.

Areas are also targeted year after year so that previous election polls and constituency data can be used to build a more general picture of voter habits.

Professor William Jennings, a polling expert from the University of Southampton and part of this year's exit poll team, told Sky News: 'We'll try just to look for patterns in that data to explain as much variation as possible so that we know that our estimates are as reliable as they can be. 

'We'll model the change in the vote at each of those polling stations, and we'll try and look for patterns in that change and also particular characteristics of constituencies that might predict change and might predict what we're seeing across the country.'

The analysts, who are currently locked securely in a bunker without access to the outside world, will look at variables such as whether the constituency voted for Brexit or the number of people who own a car. 

Since 2005, when this exit poll model was introduced, the predictions have only missed by a few seats each year

Professor Sir David Spiegelhalter, a leading statistician from the University of Cambridge, told MailOnline: 'This limited data provides estimates of the probability of different types of people changing their vote. This is the core of an exit poll.

'The probabilities of specific vote changes, combined with knowledge of the vote-shares in the last election, provide estimates of the new vote shares.'

And, since this model was introduced in 2005, it has been highly successful at predicting election results to an accuracy of just a few seats.

During the last election in 2019, the poll was incorrect by just three seats for the Conservatives – predicting 368 seats versus the actual 365. 

The poll also slightly overestimated the success of the SNP and Liberal Democrats, this meant it slightly underpredicted Labour's eventual seat count.

In 2017, meanwhile, the exit poll was off by only one seat for both the Conservatives and Labour.  

In fact, the last time an exit poll went seriously wrong was in 1992, before the media companies teamed up to produce a single exit poll.

In that year, both of the exit polls predicted a hung parliament when in reality John Major's Conservatives won a majority of 21 seats.

Although redrawn constituency boundaries could introduce some uncertainty, the exit poll is likely to produce a good impression of which party will form the next government

There are, however, a few factors which could make this year's exit poll slightly less reliable than usual.

As Dr Bunting points out, since the last election a number of constituency boundaries have been redrawn including the creation of some new constituencies.

This could make it slightly harder for the analysts to draw on previous electoral data.

Additionally, the likely size of this year's swing could introduce some level of uncertainty in certain areas.

Professor Spiegelhalter also notes that the increased prevalence of Reform UK 'increases the complexity, but does not challenge the basic methodology.'

However, experts agree that this year's exit poll will likely produce a result indicative of the broad nature of the result.

Professor Spiegelhalter says that although he wouldn't consider the results of the exit poll to be certain he would 'shall take them very seriously indeed'.

Although we won't know exactly who's won until tomorrow morning, from 10pm onwards we should have a pretty clear idea of who will form the next government. 

Read this on dailymail
  Contact Us
  Follow Us

Read the latest local and international news from trusted sources in one place.