Safe Speed Forums

Regulars will often have heard me say that 'risks' are 'over-represented' or 'underrepresented' in the crash statistics.

It's a view that I believe helps to distinguish between risks that are theoretical and risks that are real.

For example

1% to 2% of drivers are drunk according to estimates but 9% to 25% of crashes involve drunks. Therefore drunk drivers are over-represented in the crash stats by ~10:1.

2.5% to 3.5% of drivers are on the cell phone at sample sites, yet only 0.4% of fatal crashes involve anyone on a cell phone. Cell phones are under-represented in the fatal crash stats by > 6:1.

50% to 80% of drivers are 'speeding' under free flowing conditions at sample sites, yet 5% of injury crashes and 12% of fatal crashes include any vehicle speeding. Speeding is under-represented in the crash stats by ~10:1.

These ratios are pretty huge - so huge in fact that there's very little chance that they are wrong. But I want to explore what we can conclude from the observations.

Firstly I'm a little worried about the influence of other crash causes. In a hugely simplified example, suppose we have 300 fatalities a year involving 'speeding' out of 3,000. We say 50% are speeding at sample site, yet only 10% of the crashes involve speeding. So far so good. But a 'magic' policy comes along and eliminates the other 2,700 crashes and all-of-a-sudden our 10% has become 100% with no change at all in the crashes or the behaviour. In one view the behaviour is unchanged, but in another view the relative risk of speeding causing crashes HAS jumped right up because other causes of crashes have gone right down.

Secondly I'm a little worried because it all seems far too easy and far too obvious. Do we REALLY have a 'road safety industry' employing many thousands who AREN'T doing these simple sums and smelling a great big rat?

Finally I'm interested in the idea of 'running it backwards'. According to DfT inappropriate speed within the speed limit is present in some 10% of crashes. I theorise that inappropriate speed is very risky and 'must' increase crash risk by a factor of ten, therefore inappropriate speed must be present in the traffic at something like the 1% level. i.e. We know that it is rare because it is dangerous.

Comments?

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

Crikey!!!!!!!!!!!!!!

This reminds me of a little project I’m working on – I’m trying to prove that motorcycles are safer than cars! Please don’t laugh.
I’m sure I can get close, but I need to make a few assumptions, although it’s all based on data available from the DfT.

And that’s the problem with all this data, it’s pretty meaningless on it’s own. I have realised that statistics need to be applied to the statistics.

Now, I’m no mathematician, so correct me if I’m wrong, but if we look at your drunk drivers– we start by assuming the 1% to 2% figure is correct. But then we must realise that this is not 1% to 2% of ALL categories of drivers and so we don’t have a mixture of people who drink/drive. It is probably mostly young males, who are statistically more likely to crash anyway. Most ‘drunk journeys’ are biased towards short urban trips – which are the most dangerous kind of journey and these people have a relatively small amount of driving experience, again, a big risk multiplier.

On top of this, we can start eliminating great chunks out of the crash numbers. For a start, lets get rid of every crash that doesn’t happen in an urban area, late at night, and on a Friday or Saturday. Bit rash I know, but I hope you can see what I mean.

From all this, I can convince myself that drunk drivers are not actually ‘over-represented’ in the crash stats at all.

_________________
It's tricky doing nothing - you never know when you're finished

Sorry GOBy, but 'short urban trips'?!? This may be your perception, but have you ever lived out in the middle of the countryside, where all the locals drive to the local pub, have a skinful and then drive home, safe in the knowledge that there are no police for dozens of miles around to catch them?!

Yeah, you may be right, but I still think it will be a higher percentage in citys. Like I said - I need to assume something to make sense of the figures.

Wherever it is, I bet its not everywhere in equal percentages.

_________________
It's tricky doing nothing - you never know when you're finished

Is inappropriate speed recorded for any incidents where the vehicles were within the legal speed limits?

We all know that the appropriate speed varies continuously and sometimes rapidly. Can speed be categorised as grossly inappropriate, i.e. travelling at a speed such that any mistakes or unforeseen circumstances are highly likely to result in an incident, or transient inappropriate speed, i.e. failure to respond to a change in the appropriate speed. (sorry punctuationally challenged)

You would need to know, within the 50-80% of cars speeding stats, how many were travelling at an inappropriate speed, you might find a better match to the incident statistics.

Does that make any sense?

_________________
Driving fast is for a particular time and place, I can do it I just only do it occasionally because I am a gentleman.
- James May

But we're comparing the sets of {percentage of all drivers exhibiting a behaviour} to {percentage of all injury crashes that include the same behaviour}.

Since both sets are pretty much comprehensive I don't see how I could be wrong. Obviously drunk drivers and drunk driver crashes are both more common on Saturday nights, but so what.

Any genuinely risky {definition of a behaviour} MUST appear more frequently in the crash stats because that's how we properly define the risk.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

It is now, since 1st January 2005 as it happens. 2005 has been reported; we'll get 2006 figures in September.



It's hard to know about inappropriate speeds because judgements are subjective. Some examples are gross and instantly recognisable, of course - 100mph in the High Street would be pretty obvious. But someone driving at 23mph past a school might sometimes be 'inappropriate' by 1 or 2 mph.

Forum user Observer once asked me about something similar: If you were acting as instructor from the passenger seat, how often would you say: "A little bit fast just there"?. And of course the answer is prety much constantly. I think this sort of low level inappropriate speed is seriously commonplace. But I also think it's in the range where the bad outcome is likely to be braking harder than is comfortable.

At higher levels inappropriate speed constantly courts unavoidable crashes.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

Yeah, but……..if the ‘percentage of all drivers exhibiting a behaviour’ also exhibited behaviours or qualities which greatly increased their risk factor, the fact that they were ‘drunk’ may not be the sole reason for the crash ie they may even have crashed anyway. (My assumption is that if it is a young male who’s ‘drunk’, he will be 3x more likely to crash than a 55 year old who’s been drinking.)



Well, and this is where I’m struggling, if more crashes happen during a short urban trip on a Saturday night, then these people are out and about at a high risk time.

ps Please remember this is Brainstorming

_________________
It's tricky doing nothing - you never know when you're finished

We could explore total system risk on a 'representation' basis, or we could explore 'Satrday night risk' on a 'representation basis. I think we'd get the same sorts of answers either way.

One result of using this view is that we would be forced into giving strong definitions for the risky behaviour in the first place. One huge current problem is that the terms aren't properly defined. I frequently hear camera partnership people talking about 'speed related' crashes. This is 'bending the world' to fit the mindset. Such abuses mislead, rather than inform.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

I’m saying that we MUST look at total system risk (which includes Sat night risk), otherwise it’s an over simplistic view which can’t possibly exist. All the statistical information is available but it’s not linked to give the compound risk value to groups and behaviour.

To say 2% of motorists drink and then have 20% of the crashes is far too simplistic to be true. (Just like the "speed related crash" explaination!)

_________________
It's tricky doing nothing - you never know when you're finished

It might not be the perfect information on which to build a policy, but it's one hell of a foundation. That 10:1 over-representation MUST tell us that the danger is real, and that our 'risk definition' is adequate (adequate, not ideal) to identify the risky group.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

I never said it would be easy or even possible

If you apply the 85% rule against the percentage of speeders you get 7.5 - 12%, coincidence? This is stretching, or compressing, the 85% rule to apply to only those drivers above the speed limit though.

I do not think for a second that the above has any validity, it just shows that statistical values, like numerology can be made to give pretty much any answer you want.

In a previous thread about mobile phone driving you pointed out that the high incidence of my seeing drivers in poor control of their vehicle while on the phone was actually linked to noticing the phone use because of the poor driving rather than the phone causing the poor driving.

Is the speed relation being ascribed post fact in a similar way - i.e. they are just mainly bad drivers whereas drunk driving really is dangerous due to the impairment of ability.

Edit - Just re-read this and it sounds rather critical, not meant to be just throwing ideas in.

_________________
Driving fast is for a particular time and place, I can do it I just only do it occasionally because I am a gentleman.
- James May

Sure.
I was trying to explain how the 'over representation' might occur. And I've convinced myself that it's the compound risk that specifically exists within that group.

But it may be just bollocks because that's how I 'prove' bikes are safer than cars!!!

_________________
It's tricky doing nothing - you never know when you're finished