The latest polls may show the Tories' lead over Labour increasing, but there's better news for Gordon Brown in a new report by Tweetminster. The study, which attempts to predict the result of the election by tracking the most mentioned candidates and constituencies on Twitter, suggests that Labour is on course to win by a majority of 14 seats.
You may dismiss the study as hopelessly unreliable (an earlier report found that Twitter users were disproportionately liberal) but that's not really the point.
Tweetminster isn't attempting to put YouGov et al out of a job. Rather, it is investigating what correlation (if any) exists between word-of-mouth on Twitter and the election result. Whatever the relationship between the two, we will learn more about the influence of new media on the world of politics.
A similar study in Japan during last year's general election found that in about 90 per cent of constituencies the most mentioned candidate on Twitter won the seat. But that's still a 10 per cent margin of error, large enough to make the difference between a hung parliament and a healthy Tory majority.
Tweetminster's analysis of 376 British seats, based on two million tweets, suggests the following result:
Liberal Democrats: 22%
The study also found that support for the Scottish National Party is declining. and that the Greens are performing particularly well in Brighton (where their leader, Caroline Lucas, is likely to win) and Norwich South (where the deputy leader, Adrian Ramsay, is expected to run Charles Clarke close).
It also suggests that the Lib Dems are generating more support in the key south-west marginals that David Cameron needs to win to secure an overall majority.
With so much talk about this being the first "internet election", I think it'll be worth watching to see how close Tweetminster gets to the real result.