Today is Election Day in Britain and with it come the inevitable questions about exit polls. UK Exit polls are handled a bit differently than ours. In Britain, exit poll estimates of the outcome are announced with great fanfare as Big Ben strikes 10 p.m. (5 p.m. eastern time in the United States), as per the video below from BBC in 2005. While U.S. exit polls include many questions to understand why voters made their choices, U.K. exit polls are concerned with who they voted for.
The 2005 exit poll had the number of Labour seats exactly right (356), but was slightly high on the number of Conservative seats (predicted 209, actual 198) and slightly low on the number of Liberal Democrat seats (predicted 53, actual 62).
In 1992, however, the exit polls were "famously...completely wrong" (as the BBC correspondent puts it in the first video above). The exit poll forecast a "hung parliament," projecting that the Conservatives would fall well short of the majority needed to form a government. The actual result gave the Conservatives a 21-seat majority. The 1992 exit poll snafu paralleled a similar failure in pre-election polling, produced to theories about a "Spiral of Silence" or "Shy Tory" effect and led pollsters to make a host of changes in methodology that I described in my column earlier this week.
Over at our sister-site, PollingReport/UK, Anthony Wells has additional information on tonight's exit poll, which will for the first time be a joint effort of three British television networks (Well's post is also mirrored here):
It is carried out at around 130 polling stations, and they conduct about 16,500 interviews. They try and use the same polling stations at each election (though changes in wards and polling districts sometimes make it impossible) so that direct changes from the previous election can be drawn. 107 polling stations will be the same ones as last time, with an extra 23 new ones, including some new ones in LD-v-Lab seats which were previously underrepresented. Unlike US exit polls there are no questions about why people voted, it's just who they voted for.
Interviewers stop every nth person coming out the polling station, and give them a mock ballot paper to fill in, if someone refuses they are not replaced by another person. Every hour the papers are collected and phoned back to HQ, where they are weighted for differential response rates and crunched by people like John Curtice, Rob Ford, Clive Payne and Steve Fisher (if you were watching the BBC's campaign show last night, Steve was the chap demolishing the myth of bad weather helping the Tories!). The first result comes out at 10pm on the dot, with a final projection at 11pm or so.
If you want to dig even deeper, Rob Ford pointed us to this article by John Curtice and David First on the methods and models used in the 2005 exit poll.