Part 2 of a two-part guest series by Harry Hayfield. Part 1 can be found here.

Posted by their confidence at predicting Election 1997 to within an inch of the actual result, the pollsters headed into Election 2001 safe in the knowledge that they couldn’t get the result wrong no matter what the British electorate threw at them and so the BBC announced their exit poll for 2001 with it’s customary 2% margin of error. Labour 44%, Conservatives 32%, Liberal Democrats 17%. And that’s more or less what happened. The only trouble the pollsters had now was forecasting the Liberal Democrats (as the Lib Dems polled 19% and were the main gainers at Election 2001), but as they would never form a government it wouldn’t really matter would it?

Between the 2001 and 2005 General Elections a revolution in polling occurred, with companies such as Populus and YouGov using the internet to get people’s opinions and when Election 2005 was announced another sea change occurred. For the first time ever at a UK General Election, MORI (who usually polled for ITN) and NOP (who usually polled for the BBC) came together to provide the first ever joint exit poll. And the result of this exit poll? Labour 37%, Conservatives 33%, Liberal Democrats 22%.

But what was interesting was the changes in seats. Something didn’t quite add up. Sure, Labour were forecast to win 356 seats and the Conservatives 209 seats. But the Lib Dems: only 53 on a swing from Lab to Lib Dem of 4.5% and a swing from Con to Lib Dem of 2%? Were the pollsters about to get burned again by underestimating the support of the Liberal Democrats?

Well, initially that didn’t look to be the case. Sunderland South declared first (as it had done in 1997 and 2001) and recorded a swing to Con of 3.9% (slightly higher than the exit poll was suggesting) but what was this? Lab down 5%, Con up 3% and Lib Dem up 3%. That was strange to say the least and this was backed up when Cheadle (a Lib Dem gain in 2001 by 33 votes) stayed Lib Dem with a majority of 4,000!

As the night went on it was clear that a mini 1997 was happening. People were going away from Labour and voting for the person most likely to defeat Labour. This explained the SNP gain in the Western Isles (or to give it its new Gaelic name Na h-Eileannan an Iar), the Lib Dems gaining seats like Manchester Withington (on a Lab to Lib Dem swing of 17%) and the Conservatives gaining Enfield, Southgate (on a swing to Con of almost treble the national swing).

The bizarre thing is though, despite all of these radical changes, the end result of a Labour majority of 66 was the same as the exit poll! All of which poses the question with the idea of overnight counts being frowned upon following changes in how elections are regulated, will the exit poll have to be revamped yet again or will the tried and trusted method still work in the future?

Harry Hayfield is a Lib Dem activist in Wales.