Tweetminster is the website for people too stupid to work Tweetdeck or any of the many other Twitter apps. They have had problems with their headline-friendly research before – they overestimated Labour activity on Twitter by 45%. After a bit of prompting they retracted those figures in an embarrassing climbdown. Now they are claiming they can predict the election based on tweet patterns – essentially the thesis is that the more tweeted about the candidate, the more likely they are to get elected. On that basis they are predicting that Labour will get 35%, Tories 34% and LibDems 22% with others scraping 9%.
There are so many methodological errors it is hard to know where to start. The most obvious one is that the Twittersphere is not representative of the voting population. Guido wants to know, do they fancy a bet?