Friday, 8 April 2011

Some new AV polling, at last

After a nine-month gap, someone finally got round to commissioning more second-preference polling. It's a YouGov survey for Channel 4, with a large sample size, so it should be fairly accurate.

I've added the calculation to the swingometer, along with a new swingometer mode for calculating based on UNS rather than UPS swing - with the Lib Dems down as low as they currently are in the polls, the difference between the two is significant.

A few comments on the polling itself below:

  • Second preferences have a cross-tab by first preference - though there are no 1st+2nd to 3rd cross-tabs - and the sample size is large enough to see quite a difference in how the various 'Other' parties (UKIP, BNP, Greens) receive and transfer votes. While the third preference data is relatively useless, there is data on the various major party preferences for a given first preference, which makes calculating transfer rates relatively straightforward.
  • Currently first preferences and FPTP votes match up pretty well. That's not a surprise - tactical voting isn't that big an effect: where parties bother campaigning at all is bigger. If AV wins expect a slow drift in first preferences towards the minor parties, though.
  • The sample size was almost 4,000 voters. To get a decent level of accuracy on the second preferences, most polling is going to have to be this big in future.
  • They surveyed Great Britain as a whole. Really, these surveys need to be country-by-country. I've tried to adjust for this in the English swingometer as best I can.

The big implication is that projections based on polling are probably going to start needing to treat the three big 'Other' parties differently in England, rather than grouping them all as 'Other', if AV passes. I'll wait to see if it does pass before rewriting the swingometer, though.