Sunday, April 11, 2010

The Big Questions #1

As some may know, I was on The Big Questions this morning, a sort of populist Sunday morning talkshow. In essence, I was asked to debate the question "should we lower the drink-driving limit from 80mgs to 50mgs"—and to defend the status quo. I'm going to cover the experience in a few posts, for clarity, but there are a few general comments that I'd make...

First, I was incredibly nervous when it came to it. I am not usually so on-edge when doing speaking gigs; but usually I have a reasonable billing (and thus reasonable time to get a nuanced point across). Another factor was that I knew that I was being asked to defend a position that was potentially very emotive—and, sure enough, they wheeled on some couple whose son had been killed in a drink-driving accident (of which more later).

Second, although I had made extensive notes and had, I believed, a well-structured argument, this was thrown out slightly by that fact that, when the section finally came up, they had changed the question to "should we ban any drinking before driving"? Although I had anticipated this as a counter-argument to some of the points that I might raise, I had not expected to have to kick off defending that position.

Third—and this is what this post will deal with—I was ambushed by figures that I had, quite simply, never heard before. These came, most specifically, from Dr Valerie [someone or other] from the British Medical Association.

The first claim that she made was that, with 80mgs in the blood, reaction time was impaired by 12%. Sensibly, I should have asked "so what is the average reaction time, in milliseconds?" because, when I asked her afterwards, she had absolutely no idea. She waffled about lots of extenuating circumstances, blah, blah, which would probably have satisfied the audience—but her not having the figures would have put her on the back foot—as would the audience understanding that 12% of, say, 10 milliseconds is utterly insignificant.

However, the most important claim that she made was that the risk of being involved in an accident with 80mgs of alcohol in your blood (the current limit) was ten times that of someone with none.

I have found out since that this was, quite simply, a lie.

Again, talking to her afterwards, I challenged her assertion and asked her where she had got her data. Valerie had not, in fact, got the data herself (her researcher had) but the graph that she showed me was this one—and seen exactly as below.

On production of this artifact, the conversation went something like this...

"This came from the World Health Organisation. And... Well, I don't have another graph but I know that this has been replicated all over the place." She stabbed at the graph's y-axis saying, "see, there's 20.00."

"Yes," said I. "But that point at 0.8 is nowhere near ten times the likelihood of crashing."

Having noted some of the details, I have tracked down the WHO paper that it came from—the WHO Drinking And Driving, A Road Safety Manual For Decision-makers And Practitioners. Oh yes? Those discredited IPCC synthesis reports are always described as "for decision-makers"—it usually means that they are rather more political documents than nuanced science.

Anyway, you can find the graph that she was referring to in the section entitled Chapter 1: Why is a drinking and driving programme necessary [PDF, 397kb] which is not a title that fills me with the confidence that this is going to be, in any way, unbiased. For those who cannot be bothered to download the damn thing, here's how the graph appears in situ.

The first thing to understand is that this is a relative crash risk: if you drink no alcohol at all, the risk of crashing is, quite obviously, not zero—otherwise alcohol would be a factor in 100% of road crashes, rather than the 6% (2008) that it actually is.

The second point is that this is the outcome of a number of studies, starting with one in Michigan, US, in 1964, but is the one featured in the most recent of those, from 2002.
In 196 a case-control study was carried out in Michigan in the United States known as the Grand Rapids study (15). It showed that drivers who had consumed alcohol had a much higher risk of involvement in crashes than those with a zero BAC, and that this risk increased rapidly with increasing blood alcohol levels. These results were corroborated and improved upon by studies in the 1980s, 1990s and in 2002 (16–18). These studies provided the basis for setting legal blood alcohol limits and breath content limits in many countries around the world.

The real point to note is the second paragraph of the accompanying explanation... [Emphasis mine.]
The studies found that the relative risk of crash involvement starts to increase significantly at a blood alcohol concentration level of 0.0 g/dl and that at 0.10 g/100 ml the crash risk relative to a zero BAC is approximately 5, while at a BAC of 0.2 g/100 ml the crash risk is more than 1 0 times the risk relative to a zero BAC (see Figure 1.2).

In other words, at 100mgs, the risk of crashing is five times higher than at baseline. Valerie from the BMA was claiming an increase of ten times at 80mgs—which is, to say the least, a little creative.

Or, as I like to call it, a lie.

One of the things that we would obviously like to know is, roughly, what is the baseline? Presumably it is not zero, because five, or even ten, times zero is zero. So, in actual percentages, what is your average chance of crashing when you are sober? If you get in your car and drive somewhere, what is the chance that you will have an accident?

I have had an inordinate amount of trouble trying to quantify this: if I could even find an estimate for the number of road journeys made every year, that would help. If anyone knows where to find such figures, please, let me know.

Otherwise, I shall proceed to try to piece the bits together in my next post...


Anonymous said...

sorry to hear you have now got first hand experience of why facts are irrelevant in public life!

Kevin Monk said...

Lies, damn lies and statistics huh? Here's some travel safety stats. I'm not sure how useful they are to you but it might be a start...

I dug them out after I'd heard that Air Travel figures were very much skewed by which way you want to measure them. Per trip? Per kilometre? Per hour?

I'm no statistician but it seems to me that if you're at the legal limit for drinking and you live a kilometre away then you might be better off driving home than walking?? or have I just completely misinterpreted them.

Of course, none of these stats take in to account the safety of others and that includes those used by the doctor.

The real killer question is... "How safe is it to ride a horse back from the pub whilst off your tits on ecstasy?"

sconzey said...

DK, I thought you looked terrified. Handled it well though. Raging a bit about how preachy everyone was. When that girl in the red top made the fantastic point about kids; she was shouted down. That for me set the tone of the debate.

FFS; I went to church this morning and there was less dogma and more discussion and rationality in the pastor's sermon than in that "debate" show >_>

The major misapprehension that went untackled I think (that you did -- to your credit -- attempt to address) is this idea that government is a magic wand; that you can pass a law and the problem goes away.

Devil's Kitchen said...

Thanks, sconzey.


P.S. The girl in the red top was, in fact, the lovely Bella. And I thought she was rather more composed than I...!

Angry Exile said...

DK, for the chances of having a collision normally I think you're going to struggle. I was looking for this sort of thing some years ago for something else but got nowhere. The trouble is nobody seems to keep figures on damage only collisions, or if they do it's someone like the insurance industry and they keep it to themselves. You can probably get somewhere by changing your question to 'if you get in your car and drive somewhere, what is the chance that you or someone else will get injured or killed?' Numbers for deaths, serious injuries and minor injuries ought to be fairly easy to come by. If you can get figures for the alcohol related ones then one from the other should be the sober ones. Divide that by the amount of driving that goes on, which is going to be in the X billion vehicle kilometres range, and you should be getting somewhere. I suspect the first thing you'll find is that car crashes are astonishingly rare to start with. That might mean anything that has any significant impairment has quite a big effect on the numbers simply because of where the base line is, but that's just a guess.

RobH (NZ) said...

John Brignell has this piece Risk of Travel on his Number Watch website.
Also in his book Sorry Wrong Number! I am sure he mentions something about the derivation of the 0.08 limit. Comes from 1970's German crash stats on the autobahns before there were any drink driving laws - or something like that. I doubt this will be of much use to you, but you never know. (John isn't a well man these days so making contact may not be easy)

Chris said...

Ooh, will you just look at that lovingly massaged ideal sigmoid curve in the graph. If that's not evidence of cherrypicked and manipulated data (or just outright junk science) I'll eat my hat.

And re: The Big Questions. GJ arguing on a baited-and-switched position. And Medhi Hasan @55 mins - eejit!

Anonymous said...

You probably get this already but just to expand on RobH's comment: if drink driving is illegal then the people who do drink and drive are possibly more likely to crash anyway since they are the kind of people who disregard the law.

Twisted Root said...

I think the lady you are referring to is Vivienne Nathanson - a hideous hectoring old boot who calls herself a bioethisist.

There is little to be gained from these debates since; a) they are very narrowly defined, and b) you are set up to defend the indefensible and then be portrayed as basically a child killer/murderer/nazi/irresponsible wastrel/dangerous lunatic (delete as appropriate) as they wheel out an emotional victim of your wrongthink and then the highly structured debate is instantly moved on by the fast talking host the viewing audience having been guided to 'their' conclusion.

Anonymous said...

As you point out, the graph is fundamentally flawed, as the bottom line should be 1 not 0. It has been put together by someone with no understanding of statistics, so the figures in it can have no credibility. To find out the real figures you'd have to look at the underlying papers.
But - some territory is not worth trying to hold. I'd suggest you put some care into the public battles you chose to fight. Don't let the MSM label you as the bloke that wants to let us all drive round pissed.

cuthhyra said...

I would have wanted to make two points to the woman whose son died:

1) The driver was breaking the existing law, immediately suggesting that making the law even tougher would not have prevented her son dying.

2) Driving at 80mph on the wrong side of the road in a 30mph zone has NOTHING to do with being slightly over the limit and everything to do with being a complete idiot. Changing the law would not have stopped the driver being a complete idiot.

Finally, although Bella's point was sound, a less emotive example might have been something like being in a heated conversation with a passenger. Perhaps about drink driving laws.

Dick Puddlecote said...

"I have found out since that this was, quite simply, a lie."

Having extensive experience of the righteous myself, I never believed it was anything else. It jumped out of the screen as 'science by press release' stuff.

I put my thoughts on the LPUK blog, but to precis it, it's sad that whenever an opposing rational position is put in a debate such as this, the instant reaction is to fly off into hyperbole and extremes of fantasy.

One pint in a country pub with a meal and the Auto Express guy was accused of wanting people to die on the roads? FFS!

zebrastophiles said...

It was an interesting show of the power of mob yesterday and I agree with Dick above about the tendency to fly off into hyperbole.

Of far more interest to me is the way the BBC editorialised yesterday and the way the debate was set up. I hope you're gonna talk about this later.

The drink-driving debate is an interesting one. The one problem I think you're gonna have here is that through my life I have seen how one/two drinks can effect different people in different ways. Some people don't feel a thing, others are fuckered after one. Drinking affects different peeple in different ways so scientifically the effects would appear to me to be difficult to quantify.

This makes it difficult for the law makers. In an ideal world the individual should be able to know their limits and react accordingly, but I think yesterday proved that rational thought can be in short supply.

Be interested in thoughts anyways.

Anonymous said...

I watched and thought you did rather well considering that the 'panel' seemed to be made up of a selection of VERY left wing, authoritarian types.

You and the bloke from the car magazine were like Christians being thrown to the lions! Very brave.

There really is NO impartiality at the BBC. Shocking.

FlipC said...

Perhaps worth pointing out the discrepancy between the extract and the graph title. The extract states that they're discussing "crashes" the graph title "fatality". Also it doesn't make it clear if that's the fatality risk for the driver or for the crash.

As such it makes countering with stats a little more difficult, however the statistics office states that fatalities (of any involved) were 2,538 in 2008; fatalities for drink-driving accidents (again of any involved) stood at 430.

So um five times more likely to kill someone while driving sober than drunk?


Anonymous said...

"12% of, say, 10 milliseconds is utterly insignificant"


I recently had some idiot on a bicycle in the dark, no lights, dark clothes at speed (downhill) go though some lights in the opposite direction to me, while I was turning right. Missed him by 0.5 meter, say. I was stone cold sober - driving within the limit -not touched a drop. But

1. If he had been 100 ms or so later, I would have totalled him.

2. If 1. happened early morning, after a couple of wines the previous night, I'm a homicidal drunk driver. Thanks.

Anonymous said...

PS - I would guess reaction times are round the 100ms - so the 12% is 10 - 20 ms. As you say, totally irrelevant.

JN said...

Its all on the DfT website

509 billion vehicle kilometers travelled. (2008)

230,905 injuries or deaths reported to police. (2008)

Of which 13,020 (5.6%) were drink drive related injuries or deaths. (2008)

So sober injuries or deaths were 217,885.

So, for every billion kilometers travelled, there are:
428 sober injuries or deaths
26 drinking related injuries or deaths
454 in total.

These include pedestrians and cyclists.

Now I'm hungover so for some reason I can't turn that into a "if I travel 20,000 kilometers the odds of there being an injury or death in an accident are so-and-so to one" figure, I'm sure you can.

All from: for drink drive stats

and for other stats (pdf and excel links on right)

Some discussion on thinking and reaction times and stopping distances here:

microdave said...

@ Anon 11:16:00 AM - I hope you realise that under a proposed EU "Harmonising" law you would automatically have been the guilty party? EVEN if you were stone cold sober. This law basically says the smallest, or least powerful machine involved, is always innocent....

FlipC said...

So to tie JN's and my post together for those who can't be bothered.

Fatalities sober: 1/279886148 per mile
Fatalities drunk 1/1372093023 per mile

or for 20,000 miles:

1/13994 and 1/68604 respectively.

As for reaction times, for an unexpected event the times average at 1 second making it easy to calculate. Travelling at 30mph a 12% impaired driver will travel an extra 5 feet before reacting.

mongoose said...

As JN says above - loads of government "derived" stats. Basic journeys and miles in cars are available at


Brew your own.

JN said...

I've just thought we can't do a mile/kilometer comparison between sober and drunks, because we only have the total miles driven, and not the miles driven whilst drunk.

Your not less likely to be in a accident if you are drunk, those odds only give us the likelyhood of anyone who's been involved in a death being drunk.

FlipC said...

@JN - I wondered if anyone would spot that ;-)

Damn and I was hoping to get someone to try and start off a "Get sober drivers off the road" campaign; after all the statistics don't lie.

But I hope it highlights the type of slight of hand that can be pulled in interpreting data.

JN said...


Nice one :-)

So, chances of being involved in an accident causing an injury, are 230905 / 509 billion = 0.000000453644 or 1/2204370 per kilometer, but that includes the drunks.

John East said...

This analysis of statistics is interesting and worthwhile, but basing the debate on truth and reasoned argument is probably doomed to failure when your opponents are idiots, authoritarian bullies, and liars.

To evaluate where we are likely to end up with drink driving limits I would suggest following the money.

The nanny state fascists are mainly interested in robbing the average driver of as much money as they possibly can get away with. Banning too many drivers would impair their revenue stream.

The above argument suggests to me that the best regime might be something like a heavy fine for perhaps 0-30mg, an even heavier fine 30-50mg with one or two points on your licence, and a compulsory ban only above 50mg.

You can present all the evidence to the contrary until you are blue in the face, but something like my tariff above will likely evolve. It maximises revenue and (temporarily) placates the puritanical nannies.

The health and safety fascists will of course come back for more, but the politicos will suddenly start saying things like, “Enough is enough. Vote for us and we will protect the motorist” when they see that too many bans threatens their revenue stream.

Tony said...

Also worth bearing in mind that these statistics do not separate accidents where the only injuries are to the drunk driver or his passengers. If no third party is harmed then no problem, in my opinion.

FlipC said...

@JN - Yup 1/2,204,370 for all casualties and 1/200,551,615 for all fatalities per kilometre.

As an aside oopsie my original figures were based on 5.90bn not 5.09bn makes then 1/12,073 and 1/59,186 instead.

@John East - I agree, but as DK shows as an example, if you know (and remember you know) the right questions to ask you can puncture most of the quotes being offered.

People do like statistics which is why they show up in the media, sadly either they're not presented with enough information to make them meaningful, or people simply don't know how they can be manipulated.

As I hope I pointed out, imagine a teetotal advocate pointing out the 430 fatalities caused by drink driving and asking them why they don't go after the sober drivers who cause 5 times as many deaths.

It's loathsome tactics, but unless they really know what they're talking about it's a statistic they'll never be able to use again.

manc_ill_kid said...

"If you can afford to live in the country, you can afford to get a cab", I couldn't stop laughing at that.

Anonymous said...


Tough gig D.K. and I think you made a good fist of it there. It was always going to be so and I wondered why you uncharacteristically looked a tad flummoxed at the start, their shift of the goalpost explains it. Not easy is it? No amount of prep can truly prepare one for such arenas and the pre-determined editorial position was fairly blatant.

To be honest this programme is a waste of watts being barely one step up the cerebral food-chain than that execrable walking turd Matthew Wright's "The Wright Stuff" (Ch5) The Big Question is a piece of moron baiting TV in the classic tradition of Trisha, Kilroy, Springer, Wrighty et al. I despair that the notionally intelligent people who usually comprise its 'front row' can be so easily led by the nose to any pre-formed conclusion the producer requires. It's neuro-linguistic programming via the vehicle of the talk show of course. Another means by which the terms of debate on any nationally important issue can be framed by continual repetition of the officially agreed template.

The chances of adumbrating a truly unique idea on such a forum would be rare and that it would be understood and, God forbid, might be found acceptable is very unlikely. No reason not to try of course.

I despair that the likes of St. Jade's mum should be even entrusted with a vote if I'm totally honest. If a concept can't be imparted within 15 seconds then you've lost 50% of the public right there. He earns a pint in heaven who could even teach her to spell 'freedom' much less understand, appreciate and value it.

Most gratifying to hear these concepts aired publicly but it's pearls before swine chap. But that's how liberalism rules is it not? There's no point banning things like books, political parties or a free press when the greater part of the population are dumbed to the point of an irrelevance and unable to assimilate the knowledge on offer anyway. Much safer to crowd their lives with a constant stream of mundanities and inanities thus occupying every spare minute that might otherwise be reserved for private thought and the possibly seditious consequences.

Hope to see more of you on the broadcast meedja although I am probably not (by your definition) any kind of Libertarian.


J Demetriou said...

My thoughts, DK

Anonymous said...

so many cunts in that room

sushiguru said...


Just posted over at after watching the piece.

I especially loved the fact that Dr Valerie whatserpuss started off by claiming to be a libertarian. Obviously we need to do a bit more work on the definition there then...


Kevin Monk said...

I'm a fairly new convert to the libertarian way of thinking so perhaps the drinking and driving argument doesn't sit so comfortably with me as others.

I actually sympathise in part with Dr Valerie's stance in that I'd consider myself to be a libertarian and yet the concept of allowing someone to endanger another person's inalienable right to life just seems plain wrong.

I'm not intending to be antagonistic here so please bear with me as I try to understand the argument by posing this question...

Someone is playing russian roulette with your head as the target. You don't want them to do this. They get a kick out of it. They click the trigger and no bullet comes out. Have they committed a crime? There's no victim but you need to buy a new pair of pants.

You can have some fun here by swapping the "target" head with something more emotive like; your daughter, mum, dad, sister, family pet.

bella gerens said...

@ Kevin Monk:

That depends. How many chambers are in the gun? If my chance of getting shot in the face is 1 in 6, that's one thing.

But if the gun contains 60,000 chambers and only one bullet, I'd probably be a lot more sanguine. I'd have greater chance of drowning in my own bathtub.

Kevin Monk said...

For the sake of argument... There's 6 chambers in the gun.

I recognise that there's an acceptable level of risk to other people's lives caused by our actions.

Presumably my decision to drive a car at all, sober or otherwise, increases the possibility of others dying. What I'm interested in is the point at which my actions become an unacceptable risk for others. Is there a limit at all? At what point does it become unacceptable to risk another person's life? Can I experiment with a nuclear bomb in my garage and should my 8 million neighbours accept my evaluation of the risk to them when the consequences of failure is a violation of their "right to life" in epic proportions?

Ban 1 - Ban cars because driving to the shops kills people and the 0.00001% of people who die as a result is 0.00001% too many (at least that's the logical conclusion of the argument proposed by the audience of "The Big Questions") just because your too lazy to walk.

Ban 2 - Ban people from experimenting with nuclear weapons in their garage.

I rarely support any 'bans' but I'd probably support the second but not the first. So for me it's a question of where the dividing line is or would you not ban either?

bella gerens said...

@ Kevin Monk,

My point, of course, was that the probability of risk matters. As the probability of whatever you're doing killing others approaches 1, there's more justification for prohibiting that activity.

The risk of being killed in any kind of car accident in this country is nothing like 1 in 6. If it were, few people would dare to get behind the wheel. I think somebody mentioned above that it's about 1 in 60,000. Now the number show that only about 5% of those 1 in 60,000 accidents will involve alcohol, which of course reduces the probably of being killed by a driver who's consumed any alcohol.

Given, then, that the probability of being killed by a driver who's had a drink is so vanishingly tiny, the only effect of banning driving with alcohol in the blood would be to criminalise the great many people who will do it anyway, and to inconvenience everybody else who suddenly finds that the price of a glass of wine after work could easily be losing one's licence, paying a massive fine, and having a criminal record. If such a ban saved even a single life per year, I would be mightily surprised.

There will of course be those who say that to that one person, the ban will have made all the difference. But if we're prepared to accept that such draconian lengths are appropriate to save a single life per year from 'drink' driving, then suddenly we have to be prepared to accept similar bans, on even riskier activities, that presumably will save even more lives than that.

But never mind these numbers. The essential point is this. A great many types of behaviours increase the risk of car accidents. Consuming alcohol is only one of those, and as we can see, you're actually more likely to be run into by someone exhibiting one of the other risky behaviours. The laws, and the penalties, should be directed at the effects of incompetent driving, not the causes. After all, if my mother is killed in a car accident, is it really going to make me feel better to know her killer was stone-cold sober at the time? Not really. In fact, not at all, because then her killer has less fucking excuse for driving like an asshole.

Guthrum said...

It was a bearpit DK, we were being set up to fail

In the circumstances you did bloody well to hold your own

Well Done

Shug Niggurath said...

I didn't think you came across that terribly; good for you.

And if you ever appear on something with that Kilroy format again at least you've been forewarned that they'll likely throw a haymaker at you and you'll know to duck.

FlipC said...

@Kevin Monk - To tie this into drink driving allow me to change your premise slightly.

There's a known gunman stalking the streets of your home town. He appears at random, points a gun at a pedestrian, and fires it. There may or may not be a round chambered.

No-one is forcing you to walk around your town and no-one is forcing you not to either; you are accepting the risk so it's a wash correct?

Well no, the gunman is forcing you to play his game and accept the increased risk that wouldn't exist were he not doing this.

So as Bella states there's a risk simply in driving, but the drunk-driver, the distracted driver increases that risk by their own actions and are thus forcing you to accept it to.

The question, again as Bella points out, is by how much is that risk is increased. What is the threshold at which action should be taken?

The second question is whether prohibiting such action will actually decrease the risk. In this specific case remember that the reported incidents simply shows how many drivers were over the limit not that the incident was necessarily caused by the impairment. How many would still have been caused had the driver been sober?

This seems to be cherry-picking one cause-inducer simply because it can be tested. Those who drink and drive at the moment will ignore it and those who are careful will become paranoid over the alcohol content of that gateau pudding they've just had.

Mr Rob said...

I'm sure everyone is being commendably kind, but it is encouraging to see that DK is being honest about this. Only when you face up to failure can you learn from it - and this was, if judged by the impression it probably left on the majority of those who watched the programme, a failure, and an avoidable one at that.

Kevin Monk said...

Bella wrote: "As the probability of whatever you're doing killing others approaches 1, there's more justification for prohibiting that activity."

So that leads to my question - Is there a point at which the degree of negligence becomes critical and prohibition is warranted? Should you ban someone from driving when they're really pissed? 200mg? If you accept that there is a limit then how much is too much? 50mg? I believe it's a grey area and you're forced into having the state taking a judgement on what that level should be.

Whatever the current UK limit is (is it 50mg?) seems about right to me but perhaps a case can be made for that being lowered to 20mg? It's a subjective measure of the acceptable level of risk and depending upon how you view the data and perhaps your own personal experiences then your evaluation will differ.

I suppose the extremist libertarian (for want of a better term) could argue that drink driving isn't a crime but killing someone because your drunk is and this should be reflected in your sentencing.

But perhaps I shouldn't call myself a libertarian because I wouldn't support that notion. I believe there is a point where extreme negligence or the threat of violence becomes a crime in itself; even without the victim.

If I had watched this program without knowing about libertarianism in a broader sense then I would've come away with a dim view of it.

In short - I feel that arguing the toss about drink driving or incest or any other extremist position does very little for promoting the huge virtues of liberty such as private education, medicine, accreditation etc etc. Unfortunately, you're unlikely to get invited on MSM to argue any of the low hanging fruit but perhaps us libertarians should "pick our battles" more carefully.

Mr Rob said...

@Kevin Monk

Perhaps you are a classical liberal, or even a (gasp) conservative! Or perhaps you are a pragmatic libertarian - are they allowed?

I agree with pretty much all that you said there.

Kevin Monk said...

On a lighter note - I was intrigued by the slight dip towards the right hand side of the curve.

I assume this is when you become so pissed that you're as crap at driving as you'll ever get. Eventually of course you'll be so drunk that you won't be able to get in the car and therefore you'll be less of a public menace than someone who drunk nothing at all.

That graph goes "Biffra asymptotic" just out of picture.