Tuesday, April 27, 2010

Vehicle fatalities - speed impact percentages

The same statistics seem to crop up whenever someone starts discussing car accidents and fatalities as demonstrated by pil on a comment to a Shuttle letter - 30mph = 80% survival, 40mph = 80% fatality. Now while I don't expect commentors to always provide links, particularly when the system doesn't allow them to do so properly, this is one that appears quite often in official publications.

Checking through documentation the source often refers to the DfT, but they do so without any links. Dig through the DfT and you find the real source for which I've managed to find a partial link Ashton and Mackay 1979. Yes do look at that date again - 31 years ago. In fact take a closer look at the paper and you find that these aren't even national statistics they're for Birmingham which have been weighted to correspond to national statistics.

So this is the source for all those 30mph etc. which is odd as all the figures are in km/h. Next take a look at the bottom of page 3 (internal 40)

These variations in injury severity for a given impact speed suggest that variables other than impact speed are important in determining injury severity

Don't see that quoted at the same do we? Okay a little disingenuous as later we get
Contact with the car has been shown to be the main cause of non-minor injuries,
So does the bit that is quoted come from the graph on page 4 (internal 41)? Well 30mph is roughly 50km/h and 40mph is roughly 65km/h. Take a look at the bottom line of the graph and run to 40 and 65 then up until you hit the fatality line. About 50% and 85% although the latter is about right the former is hardly the 20% that an 80% survivability rate indicates. It's worse.

The key word is in the title of the graph - "Cumulative" that means accidents at that speed and under so that 85% figure includes the fatalities caused at 30mph too. There is a different figure at the top of page 3 (internal 40) which shows the true distribution. Spot the fatality curve it's that tiny bump at the bottom of the graph.

But let's look at the figures again and compare fatalities to all casualties.It's difficult with no data or numbers but at 50km/h (30mph) it seems fatalities are around 500 with all casualties around 2000 so that's a quarter or 25% at 65km/h (40mph) fatalities are around 500 but so are all casualties making it 100%  and this could indeed be the source of the 30=20% and 40=80% figures.

So this does show that the higher the impact speed the more likely a fatality is caused over a slight or serious injury and indeed once the speed exceeds 65km/h all accidents are pretty much fatal

Except as I highlighted this study was carried out in 1979 and it made some recommendations on page 10 (internal 46)
in conjunction with a 'pro-pedestrian' front structure would result in a vehicle less likely than present designs to cause serious or fatal injury to pedestrians

So all the figures shown and quoted from source are for 1979 vehicles and their shape and structure which I sincerely hope have since been altered, but it seems the government is still happy to use this as pretty much the only source they give out on this matter. If there have been further studies then why not quote those as they in turn are sure to quote this paper?

[Update 28/4 - Perhaps another thing to note is that this is using data derived from elsewhere so this paper doesn't state how exactly the speed at impact was derived. Vehicles don't carry black boxes, physical evidence such as skidmarks caused by braking depend on the model of the car, the load, the quality of the brakes, how forcefully they've been applied; so where did the figures come from?]


Orphi said...

Something that occurs to me — I don't know if it occurred to the authors of this report, but it did occur to me.

They don't take cars and mow people down with them at various speeds to see what happens. (Presumably that would be quite unethical…) What they actually do is measure the car accidents that happened to occur, and try to deduce patterns from that.

So here's a hypothesis: Maybe it's not that cars travelling at 40 tend to kill more of the people they hit. Maybe it's that anybody careless enough to be driving at 40 in a 30 zone is also the kind of careless person who's more likely to have an accident in the first place? (I don't know if they already corrected for the number of accidents in each bracket.)

Maybe these careless people are more likely to drive off and leave the victim to die? (Again, I don't know of the stats already account for this.)

And when you stop and think about how they actually get these numbers, suddenly all sorts of interesting explanations pop up.

But yeah, 30 years ago? Because car safety design definitely hasn't changed in all that time, has it?

In fairness, until recently the safety of the driver and passengers has been the primary focus; it's only recently that people started looking at pedestrian safety. And of course, even if modern cars are safer, there are still people driving cars that are 40 or 50 years old. (I think I saw one today…)

FlipC said...

Remember most studies tend to be very focussed; in this case they're looking at relationships between speed and damage not speed and accidents. So while it may well be that those who drive at 40mph in a 30 zone have more accidents that's not what's being studied.

But yes how many of those fatalities were caused by hit and runs, how many wouldn't have been fatal had the driver stopped or help arrived quickly. In this fragment there's no definition of "slight" or "serious" injuries for the first graph, which changes to "non-minor" in the second.

The bits that don't get quoted are all the caveats the authors quite rightly throw in. It's not a good scan but it's not long and worth a read. It's just the fact it's 30 year old data that'd be like if I were to try to show poor internet connection speeds in rural areas by quoting a paper written in 1985.

MERGED said...

Families of people killed in accidents 3 months ago or 30 years ago--do not care what the stats say--until you have been the parent or loved one of a person that has had that awful news--your loved one has been hit by a car--I'm sorry, she's gone. The stats mean nothing. When you look at the report and see that your child, your beautiful, intelligent, funny, loving 15 year old angel has been thrown 72 feet at 58 miles an hour--you cannot properly grasp the concept of the fact that if the moron that was driving had actually been driving the speed limit, and was actually paying attention--your daughter would be upstairs asleep in their bed, instead of in a cemetery. When municipalities start actually spending money on pedestrian safety, like sidewalks and road markings, and people start obeying the speed limits, maybe then, we can stop reading about all the stats and studies--and actually get to watch our children grow up.

FlipC said...

Sorry Merged but we have these statistics for a reason and that's to avoid this very, and understandable, emotional response.

Every time we get a debate in the media regarding this matter the parent of a child who was killed is wheeled in to confront people like myself using the facts.

Imagine if I was arguing against the subsidy of aviation fuel and the presenter brought in the CEO of an airline to cry and weep and ask why I was trying to ruin his life. Would anyone have any sympathy for them?

The loss of a loved one is a terrible thing I do not wish to belittle that in any way, but these people are obviously biased and in any other situation that bias would be noted and taken into account. But in these situations we're expected to suspend rationality and respond emotionally. If we don't we're 'heartless and cruel'.

I agree wholeheartedly that more needs to be done, but trotting out the same 30 year-old statistics doesn't help.