Fun with maps
Has there been gerrymandering in the UK?
The UK’s four boundary commissions recently released their proposals for new parliamentary constituencies: a set of new borders that will advantage the Tories both in England and in Wales. This has led to free use of the word “gerrymandering” among activists of opposing parties.
You can see why: according to Electoral Calculus, the proposed new boundaries would change the balance of the current parliament from a 12-seat majority for the Conservatives into a 34-seat majority. So have officials conspired to rig the next British general election for the Conservative party?
The answer to one question may be instructive: how weird are the shapes of our constituencies? Specifically: how long are the borders of the constituencies, given their internal areas? One must, of course, take account of hard boundaries — rivers, coasts and national borders. But that’s doable.
The point of asking this question is simple: in places where the political boundaries are fixed to meet political priorities, the constituencies tend to have peculiarly long borders as they dip in and out of communities. In countries where districts are drawn more “naturally” and with regard to communities, they tend to be shorter. You can see that most clearly in the US.
This is a map — a few years old — of a score devised by John Mackenzie of the University of Delaware, showing that index of weirdness. Green areas are places with good mutable-boundary-to-area ratios. Red areas are those with weirdly long boundaries.
Using the same colour scheme, this is my estimate of the weirdness scores — based on the current congressional boundaries — for terrible, partisan, gerrymandered North Carolina.
And here is not-gerrymandered Iowa.
So how does Britain look using the same colour scheme? Well, meh.
London (take my word for it) is all green.
The big splodge of orange in Scotland is two constituencies: the proposed “Argyll, Bute and Lochaber” and “Inverness and Skye” constituencies.
They do have long, squiggly borders. But if you draw in borough borders over the top, you can see they’re mostly following existing boundaries.
Similarly, the squiggly constituencies in the west of England look a bit odd at first sight.
But they’re often following borough borders, too. Or they’re ‘country’ seats surrounding ‘town’ seats.
There’s a big splodge of orange in the East of England, too. Newmarket is bolted onto the West Suffolk constituency.
But here’s the same map — with the county border drawn in. Newmarket is in Suffolk.
That’s basically the pattern everywhere. None of the scores for our seats are in the ‘red’ zone that marks out the worst of US partisan districting. So that’s good. But, more to the point, the Tories aren’t actually making their gains in areas of higher squiggliness.
There are lots of problems with this approach: it doesn’t rule out gerrymandering. And, in any case, a “natural” map might end up marginalising specific groups. This sort of process also needs to be used carefully in areas of geographical complexity (it is useless in archipelagos) and variable population densities.
So what does this prove? Not much. It was really an exercise by which I could test out some new mathsy software.
But we can say this: if our borders have been rigged, they’ve been done way more subtly than in the US. Not only have we appointed officials to do the job, they’ve disguised their rigging pretty well, too.
