UK Election 2015: Polls, voting systems, and other thoughts

It was billed as the most unpredictable election in a generation, and it surely delivered.  Nobody expected the Conservatives to finish with a majority government.  Nobody expected Ed Balls to lose his seat.  Nobody expected the SNP to get quite as many seats as they did.  And nobody expected Labour to go backwards in terms of seats, even despite the goings-on in Scotland.

I could go on. But the point of the matter is that the polls are wrong.

Let's put aside my translation of votes to seats, and look simply at the popular vote.  The opinion polls on Wednesday night put the two main parties within spitting distance of each other.  When the actual votes were counted, that was nowhere near the case for the two main parties, but it was in the ballpark for the smaller parties.

Vote Share according to the opinion polls

Vote Share after all constituencies were declared

So what happens if I were to plug the actual results into the poll models?  The results come slightly outside the range, but that comes down to some of the unprecedented swings we saw from Labour to the SNP in Scotland (40% swing in one riding!) and away from the LibDems to the Tories in the South West.

Prediction generated by plugging in actual 2015 votes.

Compared to the actual result.

I'm not sure what tweaks would need to be made to the model itself, since it's difficult to determine these kinds of wild swings.  This is especially the case in Scotland, where this is the first election since the SNP took off, so it's hard to see locality-by-locality how these swings vary.  Yes, there are the constituency polls that Lord Ashcroft has been doing lately, and I'll need to take a look at those, but if the national polls were off, I'm skeptical that they'd be of that much help.

A look at different voting systems

Alternative Vote

What if the UK had decided to adopt AV in the referendum four years ago?  It's difficult to say, since there are very few polls which ask about second preferences, and impossible for preferences beyond that.  But we can approximate, thanks in large part to the semi-rise of UKIP, an AV system would be the biggest help to the Tories, giving them another nine seats or so.  This majority would be more than enough to last for an entire parliament.

Result using Alternative Vote

Proportional Representation

There are many flavours of PR, so in this case, I will focus on a party list system using d'Hont's method.  Each region will get the same number of seats as present, and a party must get at least 5% in a region to get any seats.  Here, we would get a hung parliament (as you'd expect with PR), with the Conservatives as the largest party.  The Tories and UKIP would have a majority between them, and it would be difficult to find a coalition led by Labour with these figures.

Result using Proportional Representation

Other Thoughts

This is the worst result for the Liberal Democrats (or their predecessor, the Liberals) since the 1970 election.  It took 30 years for the likes of David Steel and Paddy Ashdown to build up their party to the levels they had five years ago, and it took one election cycle to completely undo it.  Their trust has gone out the window, and it will take decades to rebuild that trust.  They will need a leader who will be in it for the long haul.

Early on in the night, Labour's performance was compared to that of Michael Foot in 1983 (Thatcher's second victory, and her largest).  It didn't turn out to be quite that bad, but it's still a surprise that Labour lost seats outside of Scotland.  While the name "Tony Blair" may not be popular in the party anymore, they need to remember that he has been the only Labour leader in 40 years to win an election.

David Cameron will have to exercise far better party management than he has in the last parliament.  His majority is wafer-thin (only 12 seats), which doesn't give him any wiggle room in parliament.  John Major's government kept tearing itself apart over Europe, eventually leading to its downfall, and while the EU referendum that will be coming in 2017 will help to mitigate this, it could still be a challenge to keep both his Euroskeptic and Europhilic MPs happy.

While UKIP and the Greens didn't manage to get many MPs, they have achieved a number of second and third place finishes.  This means that, in 2020, they can focus more of their resources in those constituencies, improving their chances of getting more seats in parliament, similar to what the Liberals did in the 1980s and 1990s.

An inquiry has been launched by the British Polling Council to look into why all of the polls were so far off.  We saw similar issues in the 1992 election, and some tweaks were made then.  Pollsters will need to find what went wrong this time if they want to remain relevant.

And this leaves me with one question: When is Lord Ashdown going to eat that hat?