BBC article on cycling accidents

This is what is wrong with the government's units campaign. They failed to see that drinks tend to cluster around Friday and Saturday night. Hence people posting about statistical distributions when drunk.

I'm not sure that HGV-cyclist death patterns are best explained by the Poisson distribution, though. What are the clusters you are proposing?

Well, the general idea is that patterns can only be observed in large samples. If we're talking about 8 fatalities and 7 of those are women, or are wearing blue shoes, or are called Sheila, then the PD says that that can't be extrapolated out to a larger sample, and therefore be the basis of any probabilistic claims, because randomness clusters. It's what it does. So you say that it can't be down to chance: well, it could be.