Exit polls - what are they and how do they work?

Posted on: 29 November 2024

Exhausted campaign workers, anxious candidates, and party leaders - keenly aware of what’s at stake - eagerly await the poll's results.

Meanwhile, psephologists - both seasoned political scientists and enthusiastic amateurs - will be equally invested in what is a defining moment of suspense and drama.

Setting aside the excitement and emotional intensity of the revelation, one might reasonably ask: why do we bother with exit polls when the actual results will be known the next day or, at the latest, the day after?

The election count will begin at 9am tomorrow, with, quite accurate, tallies rolling in just a couple of hours later. By early to mid-afternoon, the first official results are typically announced.

One of the key selling points of an exit poll, compared to a standard public opinion poll or an election study, is that it captures voters’ responses immediately after they leave the polling station, while their vote is still fresh in their minds.

As such, the answers are likely to be highly accurate. Recalling one's vote, especially for lower-preference candidates, a few weeks - or even days - after an election can be challenging.

Relatedly, we can be confident that the respondent has actually voted, whereas in post-election studies, there are often significant issues with over-reporting of voter turnout.

Furthermore, exit polls, unlike opinion polls which focus on voting intentions, capture actual voting behaviour.

Overall, the information provided by exit polls is very high quality and will be used in the weeks, months and years to come to explain Irish voting behaviour.

Another big advantage of exit polls is their typically low non-response rates, as most individuals are very willing to participate when approached.

This contrasts sharply with regular opinion polls, which face growing difficulties due to high refusal rates.

Such challenges often require polling companies to apply extensive weighting and adjustments to mitigate potential biases in the data. This is not necessary with the exit poll.

A rather more practical, prosaic benefit to the use of exit polls in the Irish context is that our vote counts tend to take quite a bit longer than in most other democracies, where results are often known very shortly after the polls close.

Unusually, we don’t begin counting votes until the following day, leaving a gap in coverage that needs to be filled.

With tallies not arriving until several hours after the count commences, the exit poll offers valuable fodder for political pundits to analyse and debate during the early stages of election broadcasting.

The nuts and bolts of running an exit poll

The best exit polls replicate the voting experience as closely as possible.

In the Irish case, voters are presented with a mock ballot and asked to fill in their first three preferences.

In addition, they are typically asked two basic demographic questions (their age and gender) and a very small number of attitude-related questions, normally just three.

The whole process takes the voter less than a minute to complete, which helps keep the response rate high.

People rushing home on a Friday evening often don’t have 15 minutes to spare but they are happy to stop for 60 seconds.

Crucially, the exit polls are self-administered, to ensure voter privacy and to mitigate against social desirability bias.

When the form is completed, the respondent places it in a 'ballot box' (or in this case a 'ballot pouch').

To maximise the number of attitudinal questions that can be asked within the limited time available, the survey is divided into four 'splits'.

Each of the approximately 5,000 respondents completes the mock ballot and provides their gender and age, while the attitude-related questions rotate. Four distinct sets of three questions are asked, covering topics such as the most important issue for the voter or their views on climate change.

Who gets surveyed?

Voters are chosen on a random basis, following a 1 in n strategy, where the n is often 3. The interviewer simply counts each person exiting the polling station and asks the third person if they would be willing to participate. This random element helps ensure the sample is representative and mitigates against having to adjust the data in any significant manner.

Another critical component of exit poll planning involves selecting the polling stations where interviews will take place. In total, approximately 250 polling stations will be visited throughout the day.

Managing this operation requires significant logistical coordination, with 150 polling company employees traversing the country. Beyond selecting a representative sample of polling stations, IPSOS B&A carefully consider the timing of interviews, recognising that voters arriving at polling stations at 7am are likely to differ significantly from those rushing to cast their ballots just before the polls close.

The results are returned to polling HQ throughout the day and are very tightly guarded - only a handful of people at the polling company will have access to the data as it comes in, to avoid information leaking and potentially impacting the voting process itself. Even those involved at RTÉ, the Irish Times and TG4 will not get the results until right before they are due to be published and broadcast.

It is an intense and exciting experience for all involved but also quite nerve-racking for the pollsters, presenters and especially political scientists like me, who typically like a lot longer to mull over the data before making any pronouncements.

Do exit polls get it wrong?

Exit polls play a prominent role in showcasing the capabilities of the Irish polling industry, but they also come with considerable risk, given they are measured, very quickly, against the actual results. This risky dynamic is nicely captured in Payne's Law, named after an Oxford academic, which observes: "A bad forecast is remembered forever; good ones are soon forgotten."

While international examples of exit polls going awry are not uncommon, Ireland has been fortunate in largely avoiding such issues. The margin of error for exit polls here is typically around 1.3%, significantly lower than that of opinion polls. This is mostly due to the larger sample size but the unusually high response rates and random selection, which reduce the need for heavy weighting in the sample, also play a role.

The 2020 exit poll was remarkably accurate, causing considerable surprise by predicting a three-way tie between the major parties, each with 22% support. With a margin of error of around 1.3%, the vote shares for both Fianna Fáil and Fine Gael fell comfortably within this range. Although there was a slight under-reporting of the Sinn Féin vote, the discrepancy was minimal.

We will know in a matter of days whether the same level of accuracy has been achieved in 2024.

This article was first published on RTE.ie. See the original article on the RTE website here. 

• Tags:
• Research
• Society