Chris Hanretty, who lectures in the School of Politics, Philosophy, Language and Communication Studies at UEA, made an appearance on BBC Newsnight on 28th October to discuss his mathematical model for predicting the results of the 2015 General Election. He has been working with lecturers from Durham University and the London School of Economics to create a new website, Election Forecast UK, which will provide daily forecasts of the results right up until the election.
Hanretty considers this election to be the hardest to predict in the post-war era because of increased flexibility in voting habits and the presence of five UK-wide parties who all have a good chance of winning seats. Nonetheless, his preliminary predictions place the Conservatives with the largest share of the votes, but make Labour the largest party in a hung parliament (where no single party commands an absolute majority). Hanretty believes that no two parties will be able to join to form a majority, saying that, “At the moment our prediction is for an almighty mess”.
Speaking to Concrete, Hanretty said: “This election is the most difficult to forecast that there’s ever been, but the most likely outcome is that Labour will have the most seats in a hung parliament. It’s also quite likely that no two parties will have enough seats for a majority, which means either that we have to get used to minority government, or a three party coalition, or a second election in autumn of 2015.
“The model is based on extrapolating from current national polls to May vote shares, and on raw data from national and constituency polls which tell us how those vote shares are going to be distributed across constituencies. It’s this second part which is the tricky part”.
He said that his election predictions are made by considering how vote share at elections differs from opinion polls at this stage of the electoral cycle.
In order to predict how percentage share of votes will translate into seats, he looks at national poll samples which are broken down by constituency and determines voting patterns. Studies of the previous eight elections have shown that parties who are doing well at this stage typically lose momentum before the election and vice versa, meaning that, in his opinion, the Liberal Democrats may recover somewhat by the time the election rolls around.