Wednesday, January 21, 2015

On 20mph

Let me start this with three made up quotes.

1) (Early 2008:) We haven't had a financial crash for nearly twenty years so obviously our system of bank regulation is working.

2) Lots of Edinburgh's roads haven't had any car crashes reported to the police in the last ten years, and so must be pretty safe.

3) Drink driving should be banned, but only for people who have proved that they're dangerous drink drivers by killing someone already.

The first 'quote' could have been said by any number of our politicians just before the crash, the second is pretty reflective of a lot of arguments about road safety, and the third is hopefully pretty absurd to anyone who reads it. But they all reflect the same lack of understanding about rare events and small sets of data.

Now one of the things you learn pretty quickly when starting a career as a mathematician is that nobody likes it when you start talking about numbers at a party. Even really exciting things like this, or the fact that my new phone number is prime. But numbers are important, and sometimes you need to talk about them. There are lots of other important things too of course, like newspapers and road safety. So I thought I'd write a post about all of these things.

In particular, I'd like to talk about the Edinburgh Evening News's latest line that it has no problem with the speed reduction in principle, but that it should be restricted to roads where it will 'improve safety'. Neil Grieg, director of policy and research at the Institute of Advanced Motorists, is quoted as saying

“When Lesley Hinds and the council talk about 20mph limits, they always talk about road safety, so you would expect that one of their key ­criteria and one of their key measures of success would have to be accidents. I’m surprised they have not taken accident rates into ­account. It seems to be quite a serious omission. They should have taken crash history into account, but the problem was they were going for this blanket approach. And when you start to have this blanket approach, you are going to have 20mph limits on streets that are perfectly safe to drive at 28mph or 27mph.”

Well Neil, you shouldn't be surprised. The issue here is that you've assumed that if a road hasn't had any accidents in the last ten years then its safe. Forgive me talking maths, but what you've done is that you've failed to understand the statistics of rare events. Let me have a go at explaining.

The chance that some particular car driving down a small road in Edinburgh at 30mph will hit someone is really extremely small. It's so small in fact that some roads like Livingstone place (just next to cafe Victor Hugo) haven't had any accidents in the last ten years. But there are lots of cars which do lots of driving down lots of roads in Edinburgh, and so collectively Edinburgh's small roads host a huge number of accidents. Indeed, the road next to Livingstone place, Gladstone terrace, did have an accident in 2010 when a car hit a child. There's no real difference between Livingstone place and Gladstone terrace, it's just a quirk of small numbers. Of course, this accident was far more than a piece of data to the human beings involved.

So what the council did was to to stop fixating on the small, inadequate data, and look at the big picture. And fortunately here we have lots of data! You might like to read this review of the evidence, which shows that 'Twenty mile per hour zones and limits are effective means of improving public health via reduced accidents and injuries.' And while you're at it, you could read what Leslie Hinds and SRD have to say. And then you'd probably conclude that the council have implemented a bold, visionary policy to improve the safety and lives of their citizens, and become almost as much of a fan of the new 20mph zones as I am!


  1. Thank you. Beautifully expressed. Data are essential to debate but data can't think, ask questions, organise themselves or make decisions. All those things need intelligence, vision, reflection, knowledge, skill, experience and wisdom. Most of us don't even understand that there is a difference between "data" and "statistics".

  2. Thanks! Just spent a happy twenty minutes reading your blog too!