Tuesday, 27 May 2008

Road Injury Severity Map of England

Here's something curious. Using the Department for Transport's road accident statistics for 2006, I calculated road accident severity figures for each county in England. The DfT's accident figures include a Killed-and-Seriously-Injured (KSI) statistic, which is the count of the most serious road injuries: those leading to... well, death or serious injury. For each county, I divided the number of road accidents with a KSI outcome by the total number of recorded accidents. The reasoning was that the higher this ratio is, the more severe the outcomes of that county's road collisions tend to be and, in one sense, the more dangerous that county's roads are.

The statistics I computed gave figures ranging from just 5.6% of road accidents having KSI outcomes (in Plymouth, where it seems most road accidents tend to end okay) to 23.7% (in North Yorkshire, where almost one-quarter of all road accidents lead to someone being killed or seriously injured). I then normalized these values to lie on a scale from 1 to 100, converted these into saturation levels of the colour red and filled in the counties on a map. Yes, it took a few hours.

The map reveals a few points of interest:

* We have to work in call centres, but at least we don't get squashed too badly - Former heavy industrial areas around Manchester, Merseyside, the West Midlands and the Potteries stand out as having quite low accident severities.

* Leafy suburbs, dead bodies - the affluent south-western corner of Greater London suffers quite a lot of serious road accidents. It's so tempting to make a link to all the SUVs...

* Mad Crazy Viking Berserkers - North Yorkshire and the East Riding. One in four road accidents in North Yorkshire has a serious outcome. My father lives there. I'd like him to move away now.

* Wiltshire and Northamptonshire are really dangerous - I live in Wiltshire! Can everybody take more care please?

* Odd pockets of safety - Plymouth wins here, but there are other counties that stand out from their neighbours: Surrey, Rotherham, Newham in London.

Sadly, these statistics aren't broken down by county for Wales and Scotland. However, overall Wales is at a similar level to Southampton and Coventry, with just under 11% of road accidents having a KSI outcome. Scotland is somewhat worse, and with 17% of all road accidents ending in death or serious injury is similar to Essex and East Sussex.

My hope in doing this was that novel insights might reveal themselves if I looked at the accident data in a new way. I wondered if there would be a North/South split, or patterns that follow motorways. Perhaps the most striking thing to emerge here is the suggestion of more serious road accidents in rural areas. (Most notably, wherever there is a city that has separate figures from the county that surrounds it -- Reading in Berkshire, Leicester in Leicestershire, Poole in Dorset -- the city always has a lower accident severity score than the surrounding county.) This seems on the face of it to support the idea that higher levels of road crowding and reduced speeds are good for reducing the severity of road accidents, although we must also consider other factors specific to rural areas such as twistier roads which give poorer visibility, and possibly higher levels of drink-driving. But then we have to explain why Devon and Cornwall -- perhaps the most rural and twisty-roaded counties -- aren't particularly bad.

Like all good data explorations, this raises more questions than it answers, but I think it shows the value of looking at data in novel ways.


Anonymous said...

Excellent. One note: I discovered that DfT road accident stats don't distinguish between incidents at junctions with or without signal control. A London policeman once told me that at least half the "accidents" he attends occur at signal controlled junctions. (I put accidents in inverted commas because in my view, they are events contrived by the rules of the road.)

Anonymous said...

That's bloody brilliant! Especially the thinking about the different dangers between leafy suburbs and crazed Norsemen.

One of the 'facts' (which may or may not be true - who am I to say?) that's often trotted out is that women are better drivers than men. The logic being that they pay lower insurance premiums, therefore must be safer. What I recently heard though (again no vouching for the truth of this) is that women actually have more accidents than men, but these tend to be low speed parking prangs / scrapes, while men have fewer accidents, but these tend to be high-speed driving like a loon wipe-outs. Just the sort of thing that would lead to a KSI point in your data.

As our resident Expert On The Truth Behind Road Safety Stats, got any thoughts on this?

Ian Walker said...


You can make either men or women the 'safest' drivers depending on how you break the stats down. The first lesson in my Traffic and Transport Psychology course is all about different ways of breaking down accident figures - accidents per trip, per capita, per km, per vehicle km, per year, etc - and when each statistic is appropriate (e.g., for flying, accidents/trip is more useful than accidents/km).

If I remember rightly, women tend to have slightly more accidents per capita (i.e., in a given year, Ms X is slightly more likely to have an accident than Mr Y) but on average drive quite a bit less than men; this means men have more accidents per kilometre. For driving, this latter figure is probably more valid and men are 'worse' drivers. And this is before we even get into the issue of how severe the accidents are (I can't remember any figures off the top of my head, but your comment about lower speed collisions rings true: I think I have seen that somewhere).

The fact men are worse than women becomes completely undisputed when we consider young drivers; despite having been a young man myself once, I have to say that young men really don't belong on the road.

Chris Hutt said...

Isn't there a problem with the compilation of "accident" statistics.

At the boundary between serious and slight injuries there is not a gulf but a continuum. Judgments are made as to whether an injury is recorded as serious or slight, judgments which are unlikely to be consistent across the country.

There is also the question of the level of under reporting of slight injury accidents, which is also unlikely to be consistent over the country. So if slight injuries are less likely to be recorded in, say, urban areas the ratios will be distorted.

Anomalies like Devon and Cornwall might be explained by higher levels of reporting of slight injuries and/or lower levels of recording serious injuries due, for example, to different local attitudes.

Anonymous said...

Have you looked at the work by EuroRAP on risk mapping? http://www.eurorap.org/risk_maps

Ian Walker said...

Liz - that's really interesting. I hadn't seen those particular maps before, so thanks.

Interestingly one can buy a UK map which highlights the roads particularly dangerous to motorcyclists, which is a similar concept but for a more specific group.

Tom said...

Is that the Martin Cassini with the bizarre traffic light fixation, by any chance?

Excellent work on the maps, by the way - I'd love to know why Hammersmith & Fulham is so poor compared to similar inner city boroughs in the east, though.

David Brett said...


You have done an excellent job by sharing this information. After an accident, seeking legal redress often seems a necessary evil. Makin a personal injury claim is the natural step but the myriad of legal complications can often seem like a herculean task if you’re still trying to recover from the accident in question. There are ways to make the whole process as smooth as possible and ensure that getting reimbursed for the physical or psychological harm suffered is not as time consuming and costly as it seems. For more information visit: Personal injury compensation.