Statistics generally piss us off. This is largely because people lie with them and pretend they are doing science, but it’s also because we are too thick to work with a general scientific framework, and far too thick to admit this failing.
The cement wagon in the news ploughed through the bridge wall at almost precisely the point, and precisely the time required to mount an attack on the train (2 an hour going through). This doesn’t mean the driver did it on purpose of course. Statistical reasoning starts in observations like this, and others we might be able to make about road conditions and why an accident would be more likely at this particular spot (road surface etc.) – we have very little statistical reasoning like this about in our society (much of which can be done without the maths). The driver may be completely innocent of course, but what would we think if he repeated next week and the week after? In the real enquiry, the accident investigation will be into skid marks and so on.
What we miss in public statistics is the history of whether we are listening to ‘an innocent accident driver’ or some bastard lying to us yet again. Politicians are nearly all in the latter category. We need a form of journalism that will stop this lying by making it utterly clear to us it is happening more directly in argument than Paxman’s yawns. A different and more intelligent attitude and practice in public statistics would help. Statistically though, we are too stupid to take part!