Admiration or pity?

When a pollster tells people his work involves exit polls, there are two main reactions: admiration or pity. Non-academics lean towards the first, their typical question being “How is it possible to tell precisely the outcome of an election just after the ballots have been closed?”, while scientists tend to express the latter when they ask “How can you be so crazy as to produce a projection of an election, the outcome of which will be published just hours later?”.

In the end, both are right: Yes, it is risky to produce a projection which will be measured against reality just hours later. And yes, with only a few exceptions, pollsters forecast the outcome of elections based on exit polls very precisely. It is not a miracle; it is simply a science-based craft, the basic requirements of which I will describe below.

Before that, however, there is another question to be answered: Why would anybody spend a lot of money to find out something which he will know for nothing just hours later? The reason is that, for the clients (almost exclusively TV stations), those six hours between the closing of the ballot offices and disclosure of the first official result of an election are crucial. Audiences and political players don’t want to speculate about the outcome, they want to know. It is only because the results of an exit poll are so reliable, that politicians are willing to comment on the results and millions of viewers are willing to watch. (Viewing rates for election reporting are among the highest in most democratic states.)

That is why exit polls have become so successful the world over, since they were first carried out by CBS in the USA on the occasion of a gubernatorial contest in Kentucky in 1967.

Richard Hilmer, "Exit Polls – A Lot More than Just a Tool for Election Forecasts", in Marita Carballo & Ulf Hjelmar (Editors), Public Opinion Polling in a Globalized World, Springer-Verlag, Berlin Heidelberg, 2008, pp. 93 - 94