Forgive my pm-ing you directly. I've seen your modelling posts occasionally on pb.com.
You seem like someone comfortable with statistical methods. I have a question. I have prepared a thread myself on serial autocorrelation in UK GE results, using the Durbin-Watson test. I was wondering if you're familiar with this test at all? If so, might you have a spare half hour just to check what I've done for howlers?
Sorry to bother you anyway.