Tuesday, 5 November 2013

Alcohol's awful superpower - road toll edition

I'd missed this when it came out. From the Herald, regarding the prior Labour bill seeking to reduce the BAC limit:
The bill's policy statement said: "There is demonstrable evidence and research already available which shows enough driver impairment between the proposed 0.05 limit and the 0.08 limit to warrant action. A drug and alcohol expert from the United Kingdom has estimated that this measure could reduce our road toll by two-thirds as it would alter driver behaviour." [emphasis added]
 From the latest road crash stats:

Alcohol or drugs were contributing factors in 30% of fatal crashes and 14% of injury crashes. Note that percentages do not sum to 100% here - many things can contribute to a single crash. 

If we assume that alcohol and all other drugs disappeared, and that every crash that they contributed towards would never have happened, the best you could hope for is getting rid of 30% of fatal crashes and 14% of injury crashes.

Alcohol must have some kind of awful superpower if knocking the BAC down to .05 from .08 can get rid of 2/3 of the road toll. 

Note that this wasn't the Herald screwing up - it's a direct quote from the legislation. Now maybe you could ask just why the Herald didn't stop to think "Hey, waitaminute. Is it even plausible that alcohol's involved in 2/3 of the country's road deaths, let alone that reducing the BAC limit from .08 to .05 could cut the road toll by anywhere near that much?"

I think I understand why there's so much support for reducing the BAC to 0.05. People think that two-thirds of the road toll is caused by booze.

I wonder if Lees-Galloway would like to bet with me on next year's road toll if National goes through with the .05 limit. I'm happy to put some money on that the total 2013 road toll will not be outside the 95% confidence interval you'd get from a simple time trend on the existing data.



  1. If we assume that alcohol and all other drugs disappeared, and that every crash that they contributed towards would never have happened, the best you could hope for is getting rid of 30% of fatal crashes and 14% of injury crashes.

    I would think that you wouldn't even get that much as some/most of those trips would still need to be taken and sober people still have accidents. So say drugged/drunk drivers are 5 times more likely (number purely conjured to make point) to crash then you could remove that 30% but you'd have to add 6% back. So even if you magically removed drugs and alcohol I think you could only end up with a 25% fatal crash reduction.

  2. Surely the government should be banning loss of control?

  3. I am interested in the measurability of these accident
    For instance how do you determine from a dead man [ fatal
    accident ] if he lost control, was diverted, sleepy or asleep, aggressive, didn’t see properly, didn’t care, dodging pot holes, looking at trees, trying to scare a cyclist, worried about his kids, .. that’s right you can’t .
    Its road traffic nonsense.
    Now the measurable.
    This guy has 0.05 % alcohol in system... therefore an alcohol related death,

  4. Agree entirely, BMK. The highest possible non-utterly-insane number would have a 30% drop. The highest plausible number would be much much lower: alcohol is contributory in each of those but can hardly be causal in all.

    Agree sober people have accidents too and so base rate accidents have to be added back in. Again - was trying to point out how off-the-scale nuts Lees-Galloway's numbers were.

  5. Car black boxes that automatically signal the police in case of speeding, sustained loss of traction, or anything else suspicious.

  6. Looking at the road toll of just over 300, about 26% of deaths due to alcohol (BAC 80+)being a factor and an estimated 3.4 deaths BAC 50-80.. it looks to me that your road killer is much more likely to be sober, that the drunk driver is three times less likely to kill you and the likelihood of being killed by a driver of BAC 50-80 is miniscule.

    You know.. I think we should take a serious look at those sober drivers who kill and injure the vast majority on our roads.


  7. Yep I got that and agree with you entirely. Just annoys me how people always when dealing with statistics remove something and forgot to add in the effect of the replacement.

  8. You can easily get boggled by statistics. They are used primarily to obfuscate commonly observable facts.

    Nothing beats observation.
    Nothing beats science.
    Sadly, nothing beats intelligence.

    No intelligent female will drink more than 2 cocktails in a night.
    No intelligent male trying to seduce her, will do more.
    We have the science:

    ANY alcohol in the body is dangerous when mixed with the tired, frustrated, angry drivers out there. Add in the drivers also under the age of reason (22) plus the majority of dumb ones (fearless), so lowering the alcohol limit to zero HAS to be a positive (just one) step in reducing wasteful medical costs.
    But nothing can work without "punishment to suit the crime".
    Ask any local traffic cop. They even get to know the ones they pull up regularly for multitudes of **repeated** offences.

    As a good medical specialist friend of mine says, "we can't even measure the neural disruption of most drivers on the road any more. Almost everyone is a walking chemical cocktail, starting with the constant dosage of fluoride+chlorine+aluminium hydroxide they've ingested through their hot morning shower."