It is a recurring phenomenon in, among other things, the national House of Representatives elections: the exit poll. Before the actual results of the elections are known, the results of this exit poll will be presented with a forecast of the final results. This forecast usually comes very close to the final result: the exit polls conducted up to and including the 2017 House of Representatives elections have an average margin of error of approximately two percent. Where does this large difference in accuracy with other types of opinion polls come from? What exactly is an exit poll and based on what data is it created?
What is an exit poll?
An exit poll is an opinion poll of the voting behavior of voters and is used, for example, in the national elections for the House of Representatives of the States General. However, the big difference between standard opinion polls and an exit poll is that the latter asks which party the voter voted for, rather than which party they expect to vote for. An exit poll is therefore held on the day of the election itself and immediately after the voter has voted. This ensures a relatively accurate result that is often discussed in detail before the actual results are received.
How is an exit poll conducted and how reliable is it?
Ipsos is a research agency that is active worldwide in, among other things, conducting various opinion polls. In the 2017 House of Representatives elections, Ipsos took care of the exit poll, as in most national elections in previous years. During the 2017 House of Representatives elections, they selected approximately forty polling stations spread across the country, with a total capacity of several tens of thousands of voters. The selection of polling stations is based on personal aspects of the average voter such as social class and age, so that the results are representative for the whole of the Netherlands. The selected polling stations are located in both cities and villages. After voters have cast their vote, they complete a separate form in which they cast their vote again, but only intended for the Ipsos exit poll. Participation in the exit poll is not mandatory, but the average percentage of participants is almost always above 80%.
A research agency not only has an overview of the votes cast, but is usually also aware of some personal data of the voters. This data can be used, for example, to estimate the number of voters who have ‘switched’ to another party, or to make connections between voting behavior and aspects such as age, gender and demographic data.
The history of the exit poll
There are various stories going around about who invented the exit poll. The fact is that the current exit polls are conducted in almost the same way as in the Dutch House of Representatives elections of 1967. The then 29-year-old Marcel van Dam is a sociologist and graduated with a thesis on electoral sociology. He came up with the idea of using a sample to forecast the election results during that year’s elections. Based on the background of the voters (including gender, age, social class and education), he selected four polling stations in Utrecht whose voters should approximately represent the population of the Netherlands. In this sample, voters were asked to fill out another form after casting their vote in which they had to indicate who they had just voted for. Ultimately, approximately 85% of voters responded to this request, which gave Van Dam more than 3,400 results. The results were analyzed with computers and Van Dam personally presented the prognosis on national television that same evening. Due to the surprisingly accurate predictions, he continued to conduct the exit polls in the following years, until a research company took over the job from him. In 2020, Ipsos is the research agency that conducts all national exit polls in the Netherlands and many other European countries.
Another story about the origins of the exit poll also comes from the year 1967, but a few months later in Kentucky, United States. Warren Mitofsky, who already worked as an opinion pollster, set up a poll for the gubernatorial elections in this American state on behalf of CBS News and eventually gave it the name exit poll. He is therefore seen as a founder by various sources. In addition, it is claimed that ballot box polls (exit polls) were already held in the American city of Denver in the 1940s.
Exit polls and fraud
Exit polls are also regularly used as an additional check on the conduct of (political) elections. If the exit poll deviates significantly from the actual result, this may be an indication of fraud or system or process errors. In addition, an exit poll can also influence the conduct of the elections if the result of the exit poll is announced before all polling stations are closed. Announcement of the results of an exit poll too quickly is therefore prohibited in some countries.