Picking a Winner

(May 5, 2015)

So what will the winner of the battle of the UK General Election be able to tell us about projection modelling? I'm not talking about the parties who will gain a share of power after May 7th, but which of the polling organisations will most closely forecast the results.

The record of UK pollsters hit a memorable low in the 1992 General Election. In the week prior to the vote they were finding an average Labour lead of 2%, thus spectacularly failing to spot the actual outcome: a Tory win by almost 8%. With the benefit of hindsight, it is easy to see that their models were systematically wrong. The Market Research Society undertook a review of the debacle which identified both methodological and behavioural reasons. These…

Read more

Tags: projections, model risk

Find by key-word


Find by date


Find by tag (show all )