Will the General Election end the YouGov controversy?

With the Tories claiming that the pollsters have a systemic bias to Labour there’s going to be much more focus on the battle that has been dividing the polling world – the one between the internet pollster, YouGov and the firms that carry out conventional surveys.

A flavour of the ferocity of the argument can be seen in the comments from Andrew Copper, boss of Populus, in these exchanges here from UK Polling Report. For Populus and YouGov are showing a very different view of UK opinion. If Populus is right then Tony Blair is heading for another landslide – if not then Labour will be struggling to have an overall majority.

For every 6 Labour votes, that YouGov is predicting Populus is saying that there will be 7

For about every 9 Lib Dem votes that YouGov is suggesting Populus puts it at 7.

In deciding between the methodologies it is too easy to support the pollster that fits with your view of the world.

The traditional argument against internet polling – that many more people have phones than web access – is declining by the day as the web access proportion is now put at more than two thirds while increasing use of mobiles means that the number of landline users is going down.

A second, perhaps more difficult, argument for YouGov to deflect is that the firm only polls amongst those who have pre-registered to take part in polls in return for cash. They are self-selecting and as such, it is argued, are not representative. YouGov can point to their efforts to broaden their panel beyond the “anoraks” and, indeed, the proportion of those surveyed who say they did not vote at the last election is about the same as the mainstream firms.

The usual YouGov response is to say “Look at our record” whenever surveys have been tested against real votes. At the Scottish Parliament Election in May 2003 and the Euro Election last year their “ever of poll” predictions were the best. But the conventional pollsters will point to other surveys where YouGov’s record was not as good – the US Presidential election and the exit poll it carried out at the June 2004 Euro vote.

Who knows? Our prediction is that as we get closer to polling day the phone pollsters’ vote shares will start to converge with YouGov’s and nobody will be any the wiser after the event!

© Mike Smithson 2005