Now I have never been a fan of the Uniform National Swing (UNS); it’s a blunt instrument used to create a seat by seat view of the country if a general election were to be called. The maths and logic behind it to quickly produce a result are clever and impressive, however it has a massive flaw. This flaw is not counting for local issues.

I have has a look at one such issue in Redcar. This is the seat that contains the Corus factory that was mothballed in January 2010, costing around 1,700 jobs in the local area. Although the government can not be held directly responsible for this closure, they are going to feel some anger towards them.

This creates a pocket of feeling that is local in not necessarily felt nation wide. This results in the UNS not picking up the feeling in this area and mis-calling the seats and swings there.

Also, due to the local feelings about this plant closure and the fact that people travel more to go to work, this feeling might be felt in neighbouring seats from Redcar itself. All of this feeling is lost on the UNS and when trying to project a seat total view using it; this kind of local issue simply doesn’t show up.

This is why I’m even more interesting in the method created by us @ tmg, as this looks at local per seat feelings and this information is projected up to a seat projection and then to the national share of the vote instead of the other way around.

If you look at this table, I have shown the swing that occured in Redcar, the swing of the bordering seats and the seats that border those:

(click for full view)

As you see the swing that was recorded in Redcar is way above the regional or national swing that occurred on the day of the election and the swings recorded in all the bordering seats and the ones bordering those (bar one) were also beyond the regional & national swing that occurred.

Now it’s true that the tmg seat by seat model didn’t call Redcar correctly at the time of the election, however looking at the figures, it had it down as a seat as threat from the Lib Dems (with a very close win chance percentage between the two parties), so it was showing that this was a seat in danger for Labour. What would the UNS have shown? A safe Labour hold. The tmg seat predictor did show that Stockton South would fall to the Tories, the UNS would have put it to a marginal Labour hold.

Now this happened within a bunch of pretty safe Labour seats so even though many swings were large, they weren’t large enough to take too many safe seats from them. Imagine is this feeling / situation was replicated in South Manchester or around Edgbaston. Here you would see these kind of swing make a large difference in a seat by seat v UNS projection.

So what can I conclude? UNS should be dead or dying. Instead a more intelligent approach to polling needs to be looked into; certainly with the possible end of 2 party politics. We need to look at seat information and not national views to try and create a bigger & clearer picture of the nation. This might sound expensive, however looking at the first attempt to achieve just that by us @ tmg, it should be clear that it’s actually not expensive and indeed within our grasp. The tmg model wasn’t prefect, partly since time didn’t allow any fine tuning, however it should show that a seat by seat approach is actually the way forward and the UNS should be put to pasture as it doesn’t really tell us anything any more.