Let’s stop the clap-trap about the uniform national swing
So much tosh will be talked about the UNS (Uniform National Swing) in the coming weeks that I thought that we ought to feature a reminder of how successful it’s been in the past four election in predicting eventual seat totals from the national vote share numbers.
Thanks to Andy Cooke on the previous thread for providing this.
In 1992, Labour ourtperformed the UNS (which was about 2%) by 1-2% in marginals with Con 1st and Lab second (so equivalent to a lead of 2-4% less than the Tories actually got on the day).
In 1997, Blair outperformed UNS (of 10.2%) by 2-2.5% (depending on marginality); equivalent to a lead of 4-5% more than he actually got.
In 2001, Labourâ€™s swing against in Lab-held Con marginals against UNS (which was 1.8% against Labour) was 2-3.5% better than UNS (so a swing towards Labour, on average, in itâ€™s battlegrounds)
In 2005, the Tories beat the 3.1% UNS by about 0.5-1% on average (1-2 points more on the Tory score) against Labour in the marginals, and by 1-3% on average against the Lib Dems. The Lib Dems beat the UNS against Labour in the non-marginals, but by such an extent as to make them marginals and even capture some of them (getting an average 6-8% swing against a UNS of 4.8%).
So essentially a party that’s advanced in the polls as much as the Tories have is likely to do better in terms of seats than the UNS suggests
All of this is why I get so irritated by those alleged experts who use the UNS as though it was some fundamental law of nature. It isn’t and they are being mis-leading.
My Guardian election bet of the day is here.