Thursday, 5 May 2005

I like to watch

There was a time when I was enough of a politics junkie to watch election returns live. These days I’ll settle for reading the BBC results on the laptop while I watch the NBA playoffs on TV.

At the moment, it looks like Labour is running slightly behind its 2001 seat total (a net loss of 21 seats), with the Tories benefitting the most (+14), but there’s real no risk of Labour losing its majority unless the polls are really wrong—the current 3.2% swing to the Conservatives would have to become a 6.5% swing for Labour to lose its majority, according to the awesome Swingometer. (þ: Raffi Melkonian for the Swingometer link.)

Misdiagnosis

Apropos of the U.K. election, Stephen Bainbridge plays ‘predict the election’ and notes Labour’s massive (predicted) lead in seats isn’t matched by its lead in vote share:

It’s an interesting example, by the way, of just how skewed the British electoral system is against the Tories. If I’m right, a 3 point difference in the national polls leaves them almost 200 seats behind Labour.

The British electoral system isn’t skewed against the Tories—at least, not any more, as Scotland’s overrepresentation in Parliament has finally been done away with; it’s skewed in favor of whoever wins the plurality of the national vote. It’s almost (but not exactly) the exact same effect as we see in the U.S. electoral college: “landslides” in the electoral college are easily manufactured by relatively small differences in the popular vote.

The effect is also a partial consequence of Britain’s nonpartisan redistricting system; gerrymandering in the U.S. depresses the number of districts that are likely to “swing” from one party to the other, while the British process tends to produce a larger number of districts close to parity. (However, unlike post-Wesberry America, there is no requirement of strict population equality for constituencies in Britain.)

Shugart and Taagepera’s Seats and Votes explains things far better than I can, if you can find a copy.