Did tactical unwind lose the Lib Dems seats?

While the Liberal democrats overall gained seats (up from 51 to 62 now), they also lost 5 seats to the Tories. Was this because of “tactical unwind”?

The seats the Tories gained from the Lib Dems were:

Region	Constituency
South West	Weston-Super-Mare
	Devon West & Torridge
South East	Guildford
	Newbury
West Midlands	Ludlow

“Tactical unwind” was an idea put forward during the run up to the 2005 election. The idea was that:

1. during the 1997 and 2001 elections, large numbers of Labour supporters voted for the Lib Dems in order to beat the Tories; and large numbers of Lib Dem supporters voted Labour to the same effect.

2. in the 2005, LD and Lab supporters would be less likely to vote this way; so in areas where labour did well in 2001, they’d do worse (and the LDs better), and in areas where the LDs did well, they’d do worse and Lab better.

So, did tactical unwind happen? Let’s compare the results and swings in these constituencies
with the results and swings in the regions the constituencies are part of.

South West region

Regional share of the vote
     result  change
Lab   22.8    -3.5
Con   38.6    +0.1
LD    32.6    +1.4

Weston-Super-Mare constituency
     result  change
Lab   18.7    -1.1
Con   40.3    +1.6
LD    36.1    -3.4

Devon West & Torridge constituency
     result  change
Lab   10.2    -0.5
Con   42.7    +2.7
LD    37.2    -5.0

In both constituencies, Labour did better than they did regionally, the Lib Dems did worse, and the Tories did better.

In Devon West & Torridge particularly, the Lib Dems did very badly, reducing their share to the vote by 5%.

South East region

Regional share of the vote
     result  change
Lab   24.4    -5.0
Con   45.0    +2.1
LD    25.4    +1.7

Guildford constituency
     result  change
Lab    9.8    -3.9
Con   43.8    +2.4
LD    43.1    +0.5

Newbury constituency
     result  change
Lab    5.9    -1.0
Con   49.0    +5.5
LD    42.6    -5.6

Again, in both constituencies, Labour did better than they did regionally, the Lib Dems did worse, and the Tories did better.

In Guldford the result was very close; if the Lib Dems had done as well as the regional swing, they would have kept the seat.

In Newbury the Labour vote was already very low, and it would be hard for it to drop much further; to drop in line withg the regional swing it would have to have fallen to 1.9% of the vote, which would probably be a record low ofr a party of government. So I suspect that not much tactical unwind happened here. The main effect in Newbury was a swing from them Lib Dems to the Tories.

West Midlands region

Regional share of the vote
     result  change
Lab   38.9    -6.1
Con   34.8    +0.1
LD    18.6    +3.8

Ludlow constituency
     result  change
Lab   10.7    -2.7
Con   45.1    +5.7
LD    40.7    -2.5

Summary

In all 5 constituencies, Labour did better than they did regionally and the Lib Dems did worse. Also in all 5, the Tories did better than they did regionally. It’s therefore possible that tactical unwind did occur.

Having said that, we are looking at a deliberately selective choice of constituencies. Although we talk about a national swing, in reality in each constiturency the swing was different. In constiuencies that a party lost, it is to be expected that they would have done worse (in terms) of swing, than in constituencies they held; which may explain why the Lib Dem share of the vote in these 5 seats changed by worse than it did in their regions. Similarly, if you look just at all the seats the Tories gained from the Lib Dems, it would be expected that the Tory share of the vote rose more here than it did nationally.

So did tactical unwind happen? To answer that question fully, we’d have to do a detailed statistical analysis on all constituencies, comparing swings in Con-Lab marginals with those in Con-LD marginals. However, in the seats examined, swings happened that were unrepresentative of the country as a whole, and the Lib Dems and Tories would do well to try to understand why — the Lib Dems to reverse them, the Tories to emulate them.