“If there has been a large increase in belt wearing rates in Sweden - which there has - and if belts are effective in reducing death and injury in accidents - which they appear to be - and if one does not find a decrease in deaths and injuries following legislation … then one should find an increase in damage only accidents … .

From 1965 to 1975 the indices for car insurance claims filed in Sweden and Denmark moved up and down together. From 1970 to 1974 both countries experienced a downward trend. At the beginning of 1975 Sweden passed a law and Denmark did not, and for the first year in ten the two countries were out of phase. The graph for Sweden turned up sharply while the graph for Denmark continued to decline. One year later Denmark passed a law and its graph turned up sharply. This could be coincidence. Clearly other factors have caused the indices to move up and down in other years. It has been reported that a change in insurance policy, permitting smaller claims, was associated with the large rise in Denmark in 1970, and that this policy was gradually reversed in subsequent years (Lund 1982).However, Sweden and Denmark experienced small increases in the numbers killed in the years in which the laws were introduced (Figures 5.6 and 5.15). If one accepts that seat belts afford protection in crashes, and there is no decrease in the numbers of deaths, then it is plausible that the numbers of crashes should have increased.”

I think that risk compensation theory can be explained more simply in terms of shifting priorities. The idea that a driver thinks that now that he is buckled up he can afford to take more risks is too deterministic. The truth is much simpler. When the priority of self-preservation is fully accounted for in being buckled up, other priorities enter the driver’s decision-space. In effect, the buckled driver is at a different plain of awareness.

Whether a seat belt is buckled or not impacts on the driver’s awareness, whether consciously or subconsciously, as to what is going on around them. Situational awareness is not something confined to the jargon of fighter pilots, but relates to any situation comprising a number of variables where perception, comprehension and anticipation impact on decisions made. The best fighter pilots are those for whom situational awareness is automatic… that is, operating on a subconscious level. Driving a car is, in principle, no different. The presence of the seat belt and whether or not it is buckled is a variable that impacts on a driver’s situational awareness, and the sorts of things that they prioritize when making snap decisions in a dynamic environment. The first priority will always be self-preservation. When self-preservation is not in question (i.e., when safely buckled up), other priorities come to the fore – like meeting deadlines, “testing” driving skills, impressing friends, impressing oneself.

A buckled seat belt establishes a new limit, or bottle-neck. It encourages a driver to go where they would otherwise not contemplate. Is the subsequent increase in overall accident rates linear with anticipated risk “compensation”? Or is it a quantum jump - not “compensating” for anything, but rather, entering a different logical state - a new zone of markedly increased social “costs”, including fatalities? That is to say, will the anticipated increase in accident rates increase the number of BIG accidents, that shunt the new fatality rates to indeterminate levels? I somehow doubt that the overall fatality rate will either stay the same or decrease with compulsory seat belt laws. I actually expect overall fatality rates to INCREASE, though I cannot prove this. It’s a hunch, and that’s all it is.

Gun laws provide the best analogy for what I am trying to say. If you outlaw guns, the impact on murder rates will not be linear with anything. In fact the impact of changing the gun laws cannot be predicted, because the absence or presence of guns implies a vastly different logical state that is quite indeterminate until you actually get there. Does easy, legal access to guns increase murder rates, or decrease them? Does easy, legal access to guns reduce other crimes, like robbery, rape, etc, or does it increase them? Think carefully before providing the predictable, politically correct, anti-American knee-jerk reaction!

In the face of these sorts of questions and ambiguities, politicians should unambiguously butt out of matters confined to self risk.

We see a crash and see the outcome (injury and death being less likely if a seatbelt is worn).

What we don’t, and can’t, see is the crashes that would have happened under different circumstances but didn’t.

The only way to see those is through large scale statistical comparisons; sadly somewhat less immediate in their impact than a story of someone dying in a car crash, even if the death happened on the other side of the world several months ago.

It’s a brave politician who flies in the face of “common sense” and risks being portrayed, as John has been here, as somehow to blame for the death of every adult and child killed whilst not wearing a seatbelt.

Anecdotes like the one about the late reckless stoned unbelted New Zealander prove little — after all, my great-grandfather smoked all his adult life, and lived to be 87, but that anecdote does not prove that smoking is harmless. One has to look at such issues at the population level before the real story emerges.

I would be interested in knowing what was the response to Prof. Adams’ article on seat belt legislation in “Significance,” the journal of the Royal Statistical Society. Did the statisticians buy it, or not?