Getting a Clegg up
The striking thing about polling during this campaign is how little things have changed since the first TV debate. The subsequent debates - and even the highly publicised encounter between Gordon Brown and Gillian Duffy in Rochdale - have not made that much difference. The Lib Dems have retained most of the surge in support they gained then. Voters seem to want change, but not enough of them are convinced by the Conservatives to give the party a vote share above the mid-thirties.

At the time of writing, the outcome of the election remains in real doubt. Just about the only thing it seems safe to predict is that Labour is highly unlikely to secure a majority.

The question is whether the prospect of a hung parliament could affect how people vote. A lot depends on the sizeable number of people who say they will definitely cast a ballot but are yet to decide who for.

Under the influence
Several pollsters have been looking at the impact of the TV debates, asking whether voters changed their minds afterwards. Overall, ICM found that 17 per cent of those asked said that they had. But the results vary wildly from party to party (see chart, above right). Only 6 per cent of those now saying that they will vote Tory were influenced, compared with 33 per cent of those now planning to vote Lib Dem.

Mellow yellow
In the final debate, both David Cameron and Gordon Brown attacked Nick Clegg fiercely over his party's immigration policies - particularly the proposed amnesty for illegal immigrants. But is this an issue that could help the bigger parties to win back votes ? The polls suggests that Lib Dem supporters are much less likely to be concerned about immigration than those voting for the other two parties (see chart, below). Similarly, the EU and the euro are not key concerns for Lib Dem voters.

Bolstered pollsters
Five years ago, the results from the polling firms' final surveys all came within 1 or 2 per cent of each other. Thus, it was hard to draw conclusions about different approaches to collecting data.

But with 11 firms producing regular surveys, 2010 has been the most polled election ever. Their efforts may provide the answer to an issue that divides the industry - the effectiveness of online polls. The new internet-based firms have invariably found much smaller Labour shares and much greater support for "others" than the traditional phone pollsters.

A sharp exit
As voting closes on 6 May the results of what is surely the most eagerly awaited and expensive of polls will be announced at 10pm on the BBC, ITV, and Sky News. This exit poll will be based on 16-17,000 interviews carried out by NOP and Ipsos MORI at more than 120 polling stations throughout Britain.

Most of the chosen locations have been polled at previous general elections; a key part of the analysis will be a comparison with the 2005 findings. The academics processing the data will produce a projection of the likely result in terms of seats rather than the share of the vote. They have a lot to live up to. The 2005 exit poll projected a 66-seat Labour majority - which is exactly what happened.

Mike Smithson is the editor of politicalbetting.com