Wednesday, 23 May 2007

Risk homeostasis: How simple! How obvious! How... worrying

"Risk Homeostasis". I encountered the concept today when glancing at a helmet usage thread on a motorbike website. The thread itself was, to put it mildly, worrying. Without trying very hard I spotted people using personal anecdotes as proofs, gut feelings as proofs, cherry picking, ignoring alternative explanations which didn't fit their conclusions... You name it. And that was without looking at the articles they referenced.

You and I both know, dear reader, that statistics can be made to prove anything, and so can research papers. The trick, in the absence of personal in-depth knowledge of the issues involved or an expert you are prepared to trust, is to read between the lines, spot weasel words, evaluate what you have got the time and access and care factor to read, and make your decisions based upon informed understanding of the scientific/mathematical/psychological/(insert your specialisation here) principles at work.

For a start, and just as an aside, the phrase "intelligent people can surely see..." should make you run screaming. It usually means "people who agree with me." It's like praising a woman's intelligence as a pick-up line, in the hopes they won't try and prove you wrong by walking away. It's cheap, inherently nasty and says really bad things about your personality, not to mention your skills at debating.

Risk Homeostasis was referenced in this thread as, well, a godlike theory. So I had to look it up on the wiki. And I can see why it's popular: It's controversial enough to appeal to people who distrust the establishment or indications of nanny-statery, it's simple enough to understand and it seems clear, and clean, and passes the sniff test for "sounds okay". And that's deeply, deeply worrying because it's incomplete and horribly misleading and probably, in fact definitely, dangerous when put into the hands of those with a barrow to push. It's not so much wrong as... well, it's not even wrong, to quote a greater thinker than myself.

Risk homeostasis, for those of you who haven't encountered this before, is the psychological theory that people adjust their behaviour to suit a perceived level of risk at which that individual is comfortable, and that increased safety measures reduce the perceived level of risk, so risk-taking behaviour is increased until the same, acceptable level of risk is achieved again. Think about your own behaviour, about what you would do if your car had the seat belts removed, or if people around you are driving riskily, or if the floor is slippery, and you can see that it more or less works. Funnily enough, your brain takes in danger cues from the environment and enacts self-protective measures accordingly. But I get ahead of myself.

It argues that seat belts and crumple zones, airbags and ABS brakes don't change road trauma statistics because people rely on them and drive more dangerously. It argues that accident statistics in skydiving are remarkably stable because improved safety gear leads to increased risk-taking because it's now feasible. It's a cost-benefit economic type of theory. It's also, incidentally, a theory of unconscious behavioural change, so don't complain that you don't make that decision: That's not the point. Most of human behaviour is unconscious, at least on some level.

And Wilde, the ex-pat Dutch psychologist who created this theory, is extensively published and does a lot of research. Which brings to mind the faint echo of "methinks he doth protest too much".

Critiques of the theory have been published for a long time, and even a simple Google search will lead you to some very nice ones, so I won't bother to discuss them. Instead, I'll look at what holes I can poke without doing research (dangerous, I know).

Where it raises my hackles is threefold:

Firstly, my study in psychology brought me to the conclusion, based upon continual presentation of evidence, that neat, simple theories that attempt to explain broad swathes of anything psychology are invariably incomplete. True in some instances, possibly. Part of the solution, possibly. But everything? Anybody who is prepared to accept a simple explanation for anything as complex as "risk-taking behaviour" in humans in general probably hasn't understood the whole picture. This theory is like attempting to reduce evolution to the statement "that which survives to breed, survives to breed." Yes, but... Look at the issue of safety devices in cars leading to increased risk-taking behaviour. If we consider seat belts and airbags, stating that people are more likely to take greater risks assumes that people are willing to accept crashing because they're more likely to survive it. Really? Crash my nice brand new BMW with Automatic Stability Control and 7 air-bags? No way! What, I ask, about the increased capabilities (engine, suspension, tyres, brakes) of newer cars with increased safety features without an additional increase in driver ability?

Secondly, behavioural: Wilde argues that penalties don't change behaviour because they don't necessarily change perceptions of risk. This is not so much wrong as useless. There are less controversial, more constructive, better supported and ultimately more useful theories of human behaviour which explain why punishment does not enact lasting behavioural change, why spoon-feeding people information does not enact lasting behavioural change and why the only thing which does enact lasting behavioural change is motivation. If a person can't see a reason why they, personally, should change their behaviour there is habit, there is defiance of external authority, there are various self-talk cognitive explanations for why nothing will change unless (by defiance or contrariness) the change is opposite to the desired direction. Give them a motivation, and change will be pursued. External rewards (money, goods, privileges, expressions of love) can be a motivation, if the reward outweighs the inconvenience. Internal rewards (satisfaction, relief, happiness) can work as well, for the same reasons. Not wanting to go to gaol after the next arrest is a motivation. More money is a motivation. Wanting to keep your licence is a motivation. Not losing your current relationship is a motivation. The key theories of Applied Behavioural Analysis and Positive Behavioural Support (which really are god-like theories for management of challenging behaviours in those with behavioural issues and/or cognitive deficits) work on exactly that principle, and work with phenomenal success. Risk homeostasis adds almost nothing to this debate. It's a bit like string-theory in that respect.

The third reason is: It muddies the waters. Lets look at motorbike helmets: Opponents say that if you don't wear a helmet, you're a safer rider. Fair enough. Safer enough? Doubtful. And unfortunately, the evidence is against you. But is this really relevant? Waving aside for a moment the apparent willingness to crash (see above), no amount of safe riding can avoid crashes absolutely. I shan't name names, but one particular anti-helmet proponent who is also, even more worryingly, of the belief that a full-face can break your neck (short rebuttal: If an impact to your helmet will break your neck, the same impact to your chin will break your neck and your chin), has written in national magazines of instances where caravans or truck batteries have become detached from vehicles in front of him. This is outside your control, only your reaction to it is in your control. Therefore, you may crash. And when you crash, you are probably going to hit your head. And when you hit your unprotected head, you are going to at the very least get minor concussion, but you stand a much greater risk of a serious, innately incurable brain-injury. From a public health, disability services, national economy perspective the effect of helmets upon the accident rate pales into insignificance when compared to the effect of helmets upon the serious injury and long-term disability rates. And if you crash without wearing a helmet that's your choice, mate, but if you expect my tax dollars to pay for your $150,000/year (AUD) care costs you bring me into the equation, and I won't be happy at all.

I challenge you to look for more examples of risk homeostasis being raised in debate, and see why. I suspect it'll be in defence of a position which the facts make shaky or, alternatively, won't be of much benefit to the debate.

I welcome critiques, criticisms, advice and evidence. Swear at me and I'll probably discount everything else you say. Raise a point I've already dealt with without rebutting my rebuttal, and I probably won't even bother answering.


EoR said...

There's a similar argument about bicycle helmets. I've seen it argued that bicycle helmets make you more likely to have an accident due to the same effect.

Horseriding helmets are also contentious. There are various "proofs" against their use. They only protect your cranium (though a broken cranium is more dangerous than a broken arm). Children are more likely to have falls ("likely", not "only"). They actually cause broken necks in falls (I've not see the proof for this). My horse has never done anything wrong/I'm too good a rider to fall off (both of which lack the necessary modifier "yet"). It's my choice whether I wear a helmet or not (indeed and, if death was the only result of a helmetless fall, I'd not have a problem with that - as you've pointed out, brain damaged individuals are a severe burden on the state, as well as relatives financially, physically and emotionally).

While it's been said that the only safety device a car needs is a long sharp spike sticking straight out of the steering wheel, it would also be helpful for the anti-safety campaigners to take a tour of any hospital's brain damaged ward.

Dubito said...

I think the argument with bicycle helmets is essentially the same and I've encountered it for far longer, having been a keen cyclist from quite young, but with the charming twist of a recent report claiming that if you don't wear a helmet, motorists are less likely to run into you. I haven't read the report and can't say how the behaviour of the cyclists may have changed.

As for horses: I have spoken to the state-wide Queensland Paediatric Rehabilitation Service concerning brain injury in rural areas, and falls from horses are common enough to have been mentioned in a two-minute response to a one-sentence question.

The broken neck issue is particularly interesting and particularly contentious, with some particularly vocal advocates against full-face motorbike helmets claiming to know friends who have had their necks broken by their "mandated passive safety devices".

But the sad truth is: If your neck was going to be broken by an impact to the helmet, an impact to your chin will still break your neck and your jaw as well. If you're interested, the logistic regression model research is:

Lin M-R, Tsauo J-Y, Hwang H-F, Chen C-Y, Tsai L-W, Chiu W-T: Relation between Motorcycle Helmet Use and Cervical Spinal Cord Injury. Neuroepidemiology 2004;23:269-274 (DOI: 10.1159/000080091)

(I haven't actually read that report: I'm relying upon the research conducted by my colleague).

And yes, I wish more people would tour the facilities I have visited and seen how dehumanising and appalling the results of a serious brain injury can be.

As an aside, a motorbike magazine I picked up in England in December contained many letters about a journalist who had crashed while wearing normal jeans and suffered serious injuries that leathers or kevlar jeans would have prevented. The gist of the letters was "If you were in industry you would be prosecuted for failing to take adequate safety measures", "Oh, so you think your freedom of choice makes it okay to use up my public tax dollars, do you?" and so on. I found those letters deeply encouraging to read.

Search This Blog