Election forecasts

Predicting the result: three websites to watch

It’s clear we’re going to be spoiled between now and May with the greatest number of polls and forecasts ever published in a British general election campaign.

Among all the rogue snapshots and speculation, however, we’re already being blessed with a handful of predictions based less on instinct and trends and more on formulae and calculations.

These are well worth keeping an eye on, not least as a corrective to my own monthly forecasts (which to be fair I don’t claim to be anything more than informed guesswork).

Here are three of them:

1. Manchester University Polling Observatory

This is a monthly forecast by a group of academics: Rob Ford of Manchester University, Will Jennings of Southampton University, Mark Pickup of Simon Fraser University and Christopher Wlezien from Temple University.

They are using past polling evidence along with current polling trends as a guide to predict the likely levels of support for each party in 2015. It’s a persuasive idea: rather than simply extrapolating forward, they look backwards to find recurring models of behaviour in the 12 months before a general election.

Their most recent forecast is that the 2015 result is too close to call, with a Labour lead of 0.7 percentage points being “far too small to be statistically meaningful at this stage”. They also make the important point that this uncertainty is based on the fact that “Labour are holding their support, where the historical record suggests we should be expecting declines at this point.”

August’s forecast should be out in a few days.

2. Elections etc.

A blog by Stephen Fisher of Oxford University and researcher Jonathan Jones. Their weekly forecasts are based on the principle of statistical regression: in other words, there is always a degree of reversion to a norm, or a long-term average, in parties’ poll ratings as a general election nears. They cite three historical tendencies:
– that governments are more likely to recover and oppositions fall back;
– that parties move back towards their long-run average level of support and/or the level of support at the previous election; and
– that the Conservatives tend to over perform and Labour to under perform their vote intention figures in the polls when it comes to election day.

Their latest prediction is for a hung parliament with the Tories the largest party on 303 seats. They also suggest percentage probabilities of outcomes. At the time of writing a Tory majority is 30% probable, a Labour majority 21%, and some kind of hung parliament 49%.

3. Election Forecast UK

A brand new website that promises daily forecasts of the result. It’s the work of Chris Hanretty of the University of East Anglia, Benjamin Lauderdale of the LSE and Nick Vivyan of Durham University. They use a variety of sources to calculate their predictions, including data going back to the election of 1979, raw data from current polls and constituency characteristics.

At the time of writing they are forecasting a hung parliament with Labour the largest party on 306 seats, but they also give the likely range of variation – hence Labour could be as low as 247 or as high as 356.

I’m sure other sites will join these three in the coming months to offer regular forecasts. In fact, I’d be disappointed if there weren’t.


3 responses to ‘Predicting the result: three websites to watch

  1. Oh, and I like the cartogram at http://electionforecast.co.uk/ Nice not to be mislead by the varying geographical area of the constituencies, and it clearly makes the point about the Conservatives winning the shires in the south and east and Labour winning in central London, the north and west, Wales and Scotland.

    They have Lucas probably losing her seat (“Greens … Seat loss very likely”).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s