UK - Swings Settling Down

[DistingFlyer]

After how many declarations on election night does the size of the national swing become apparent?

Looking at 1970 and 1979-1997, I've calculated what the average swings were in each election after 10, 15, 20, 25 and 40 results and compared them to the final result:



The first set of figures use a swing calculated by adding up the votes (thereby giving more weight to larger or fast-growing seats, thereby increasing the Tory swing), while the second set calculates the swing by averaging the seats on an equal basis (usually improving the Labour swing). The first method seems to give a better result, with the exception of 1970.

Looking at the figures one can see that the result settles down quite quickly: 1979, 1983 and 1987 all got within 1% of the national swing after ten seats had declared, while 1997 reached it within 15 and 1992 within twenty.

As I mentioned above, 1970 is the exception: the Wolverhampton seats, with 9% swings, declared early, but also a number of seats that had had by-elections. The first forty seats were also all in England (which recorded a 5.1% swing), which also accounts for some of the difference. Nonetheless it is rather ironic that the election which had the most even nationwide swing of all the ones examined here was the slowest to settle down to the final figure, while the elections with the most uneven (1979 & 1987) were the fastest!

Using the second method here produces a more accurate result as the inner-city areas with falling populations tended to swing less heavily; this is also the reason why the 1979 Tory swing becomes so much less using this method.

A further point of observation is that in 1987 and 1992 the BBC forecast the Tories doing much worse than they eventually did, and in both cases (but especially 1992) the projection took a long time to get anywhere near the final result despite the very low early swings. (To be fair, they were pretty close in their forecasts of 1979, 1983 & 1997.)
Contrast this with 1970, when a Labour victory was also forecast but the commentators quickly reversed themselves after just a few results had come in.

[Phony Moderate]

Funnily enough, Birmingham Edgbaston was supposed to be one of the first declarers in 2010. That would've fuelled 1992 talk for a few minutes.

Interesting stuff though.

[DistingFlyer]

To add to this, here's a projection of gains - based on swing after 10/15/20/25/40 results - against what actually happened. Only Tory-Labour gains are taken into account here.



One can see just how efficiently Labour handled its marginals (and targets) in 1979, 1992 & 1997: although early swings were less than the final result in 1979, Labour still lost fewer seats than the figures would suggest. Conversely, by the time forty seats had declared in 1992 the Labour swing was exaggerated, yet they still made many more gains than the figures indicated. The 1997 Labour swing was slightly low in the early results, but again with fewer gains indicated.

A uniform swing of 12.5% would have been required to get 144 gains in 1997, and in 1992 a swing of 3.6% would get 39 gains. 1979 isn't quite as far off, with a 4.8% swing required to make 52 gains (but that 0.5% difference was a big one: seven more seats on a uniform swing).

Compare those years with 1970, 1983 & 1987: 75 net gains in 1970 required a uniform 4.9% swing, 1983 required a 3.9% swing to gain 44 seats and 1987 required 1.7% to get 17 net gains - on all three occasions, almost exactly what the real swing was.

[DistingFlyer]

Funnily enough, Birmingham Edgbaston was supposed to be one of the first declarers in 2010. That would've fuelled 1992 talk for a few minutes.

Interesting stuff though.

Interestingly, the swing in Edgbaston was exactly the same (1.3%) as Basildon in 1992. I'm sure the reactions of the two parties would have been about the same (though mirrored) if it had declared early.

[You kip if you want to...]

Funnily enough, Birmingham Edgbaston was supposed to be one of the first declarers in 2010. That would've fuelled 1992 talk for a few minutes.

Interesting stuff though.

I don't remember it causing talk of 1992, but I seem to remember it throwing the BBC studio off for a few minutes.