Thursday, January 19, 2012

Roads vs. Streets: Wherein the greater danger?

Michael Blastland plays around with some statistics, usefully!, on roads vs. streets when it comes to accidents and safety  in this article that appeared in today's BBC magazine. (Click here for his article in full and here for the  often quite stinging comments that it has triggered.)  Ours here is quite another focus, but it is interesting to keep our eyes open for short pieces like this.

Go Figure: Are country roads more dangerous than city roads?

- Michael Blastland, BBC News Magazine, 19 Jan. 2012

Is city driving more dangerous than country driving? It's a much harder question than you think, writes Michael Blastland.

Which is more dangerous, a bumper-to-bumper megalopolis like London, or the freedom of the open country road?

Instinct tells us that more cars equal more hazards. When roads are full, people jump the lights; when it's a scrap for the gaps, margins of safety fall; when the going's slow, half the faces at the wheel are in a map, mobile phone or crisp bag, or nodding off.

Busy places are more dangerous places. Aren't they?

It's the kind of simple question people want answered. So this week's Go Figure asks how you might go about finding out. I don't know too many people who want to dedicate their lives to the answer, so I gave it an hour to see where I could get. Here goes.

We start by saying that we can't rely on instinct. As Sherlock Holmes says in the Legend of the Copper Beeches: "Data! Data! Data!... I can't make bricks without clay."

Bad news. The Department of Transport told me that it can't say whether more traffic on its own causes accidents, they've not researched it. The police usually record contributory causes of an accident but they don't include congestion or traffic density.

But maybe this helps. The BBC's recent accident map shows every accident right down to specific roads or junctions. And it looks as if most accidents are in cities at the most congested times of day.

Maybe the most dangerous roads have good accident records and appear safe because cyclists and pedestrians avoid them.”

So I did a crude comparison of some numbers for accidents in London and that mostly rural haven, Northumberland. It turns out that London is about 19 times worse.

The Department of Transport has figures for the number killed or seriously injured in 2010.

Greater London: 2889.Northumberland: 151.

Well, sort of. Because you'll already be objecting that London has oodles more people, so what do we expect?

What we most want to know is not the number of accidents, but the rate, relative to the volume of traffic. That's a better description of the chance that any one of us will come to grief.

OK Watson, so far so elementary. Here are the accident rates for those killed or seriously injured per billion vehicle miles.

Greater London: 145. Northumberland: 90.

London is still worse. So it's true, dense traffic does cause more accidents. Although now London looks a bit under twice as bad, rather than 19 times as bad.

Well, sort of, again. Because here's the next problem - what's the right way to measure the accident rate?

We've just done it by adjusting the number of serious accidents for miles travelled. But a moment's reflection about big cities tells us that we can spend a long time driving not very far.

In other words, isn't it time at the wheel, not miles, that best measures how much driving goes on?

According to data here, traffic in Greater London on A-roads averages about 16mph, and in Northumberland about 36mph. If we use this as a rough ratio of speeds on roads in general - a crude assumption, but it will do for now - then we can do a back of an envelope calculation.

Deaths or serious injuries per 10 million hours of driving.
Greater London: 23. Northumberland: 32.

So an average hour on the roads of Northumberland is very roughly 40% more dangerous than an hour in congested London. This leads to a conclusion that sounds weird but is unremarkable - that places with more accidents can be safer.

But we're not there yet. Because the Department of Transport also has figures by type of road, and says the most dangerous is the single carriageway A-road, of which Northumberland certainly has a few.

So now we have a new problem, Holmes. How do we know if the road type is swamping any effect from traffic density?

To put it another way, what would be the accident rate on London's A-roads if you filled them with Northumberland's density of traffic?

We can do one more thing before the stats get properly serious. That's to look up mortality rates from land transport accidents, per 100,000 people.

London: 2.77.Northumberland: 4.46.

OK, all this is a long way short of proof that traffic density alone causes these differences. It tells us a lot, but it doesn't tell us that. It's a selective sample of serious accidents, not all accidents, in just two places. The calculations are rough and ready.

Is heavy traffic really protective in London, maybe because it slows us down? Is lighter traffic more dangerous in Northumberland, maybe because we're less careful? Or do the narrow stretches of A1 north of Newcastle outweigh all these factors? Or maybe London has better drivers. But let's not start that one.

This takes us into serious crunching - and we have to look to people who've studied the problem on a big scale. They too tend to suggest that there's less chance of an accident, especially a serious one, in heavy traffic.

Take this one, for example, which says: "Incidence rates involving property damage-only crashes and injury-crashes are highest when traffic is lightest."

And this one says: "Probably a higher traffic density leads to a shift towards less severe injuries."

But you might want to go further, and split the accidents by car, pedestrian, cyclist, motorcyclist and so on. And that raises another problem, described by John Adams in his book Risk. Maybe the most dangerous roads have good accident records and appear safe because cyclists and pedestrians avoid them.

You can measure a lot, and we do, and it's easy to find the data - the number of crashes; crashes adjusted for traffic - in several varieties; crashes adjusted for road type, the transport mortality rate, etc.

But how well can we really measure danger?

# # #

About the author:

Michael Blastland was born in Glasgow. A journalist all his professional life, he started on weekly newspapers before moving to the BBC where he makes current affairs programmes for Radio 4, such as Analysis, More or Less and the historical series Why Did We Do That? He lives in Hertfordshire, often with his daughter Cait, less often and less quietly with his son Joe, when he’s at home. (From Profile Books.)

# # #

Print this article