However, there are two main points I want to bring up from reading Thevoz and Baston:
The first is a general one about their data, where I’m heartened to see that their analysis of the result is based on changes in the actual numbers of votes received, rather than shifts in the percentage shares. I’ve argued before that turnout is a crucial factor often ignored in British elections, and coupled with that is the effect of shifts to and from not voting, as well between parties. Using percentages often carries with it the assumption that the people voting in this election are the same as the people who voted in the previous one, which I think leads to some lazy analysis.
I think it also – though it’s not something highlighted in this case – helps to show why local government elections and Parliamentary by-elections aren’t always a good indicator of how general elections will go, because you can’t assume the smaller sample at the former are indicative of how the larger sample at the latter will vote. I think that was especially the case this time and looking at high-profile by-elections helps to show it. Mike Thornton got 13,342 votes in the Eastleigh by-election and won, then got a small increase to 14,317 votes in the general election and lost because the Conservatives added 13,000 votes between the two The overall increase in turnout between the two? About 14,000 votes. Similarly for UKIP, Mark Reckless got 16,867 votes in the by-election and 16,009 votes in the general, that small shift downwards eclipsed by the 10,000 extra votes Kelly Tolhurst got for the Conservatives in the general election. Similarly, Douglas Carswell’s position in Clacton looked a lot less secure when 7,500 extra Tory voters turned out at the general election.
One final point on turnout: the graphs show, perhaps even more impressively than the swingometers, the scale of the SNP’s achievement in Scotland and how it was heavily driven by persuading non-voters to come out and vote for them. Again, only reporting percentages hides some of the true picture, particularly the unionist tactical voting that’s likely behind the increase in the Lib Dem vote in some of those seats.
The second main point is that there isn’t a consistent story to tell about what happened to the Lib Dem voters. There’s a degree of tactical unwind as Green and Labour votes go up, there’s a loss of the anti-system vote to UKIP and Green as well as a shift to the Tories which could either be a coalition detoxification effect or because of Project Fear driving voters who didn’t want to see Miliband in Number 10 towards the Tories. I expect there’s also a strong element of former Lib Dems staying at home, somewhat hidden by a number of former non-voters coming out to vote for UKIP. There does also seem to be in some seats an amount of ‘soft Tory’ tactical voting for Liberal Democrats to keep Labour out in some seats, though it’s hard to tell the extent of it as some of the drops in the Tory vote (especially in ‘safe’ Lib Dem seats) may be Tory voters taking the opportunity to protest vote for UKIP. However, it doesn’t appear to be on anything like the scale of the Lib-Lab tactical voting we’ve seen over the past two decades.
This is an important factor both in explaining the 2015 result and in looking at the strategic options for the Liberal Democrats going forward. One interesting book on electoral theory I’ve been reading recently is Gary Cox’s Making Votes Count which looks at how voters strategically co-ordinate their votes for maximum effectiveness. One example of this is his application of Duverger’s law, and the way it structures the vote within constituencies so that they tend to become two-party contests in single member plurality (‘first past the post’) elections. (Duverger is often taken to apply solely at the national level, but Cox points out that his work is just as, if not more, relevant at the constituency level)
I should probably write a longer post specifically on Cox in the future, but the important point he makes is that winning individual elections is a co-ordination problem for both parties and voters: the latter trying to determine who are the potential victors, the former trying to work out how to position themselves as a potential victor. However, the key point here is that even if a party can show that it is one of the potential victors, it can only attract tactical votes from those who won’t win if those voters can perceive a relevant difference between the two potentially victorious parties. Thus, it’s hard to get a hardcore UKIP voter to tactically vote Tory to keep Labour out because both parties are part of the ‘LibLabCon‘ they despise, and it was hard this year to persuade Labour and Green voters to vote Lib Dem to stop the Tories when they saw no difference between the two parties. Because the non-Tory vote was heavily fractured and generally not co-ordinated, that allowed the Tories to win a number of seats with relatively small shares of the vote – as Thevoz and Baston point out, many Tory gains from Lib Dems were with smaller numbers of votes than had won the seat in 2010 because of this effect.
There’s a good news and bad news conclusion to this. The good news is as Thevoz and Baston say: the Tory majorities in a lot of the seats they gained from the Lib Dems aren’t overwhelmingly massive and impossible to overwhelm in the future, but the bad news is that the only way those seats can be won back is by convincing non-Tory voters that not only are the Lib Dems capable of challenging the Tories in those seats, but that there’s reason for those voters to believe there’s a sufficient enough difference between us and the Tories to make it worth their while shifting. That part isn’t as simple as it sounds, because it’s not just about the messages Lib Dems put out, but how much they co-ordinate or clash with the messages coming from the other parties and the media generally. It’s one thing to persuade the sort of person who turns out at a local council by-election that it’s OK to vote Liberal Democrat again, but how do you get that message over to rest of the electorate?