In most countries arguments about seat belt legislation are dead. But it remains a live issue in the United States where such laws are a matter for individual states. As a consequence there exists in the United States a variety of laws and levels of enforcement, and considerable debate about their effectiveness and moral legitimacy.
A recent article on the subject in Time Magazine (“The Hidden Danger of Seat Belts”, 30 November 2006) cited research of mine done many years ago that concluded that seat belt laws had been ineffective in all jurisdictions that had implemented them. It provoked a number of hits on my blog and inquiring emails – hence this blog which attempts to answer some of them.
Why, if I was right, did so few people know that seat belt laws were ineffective? And why had so many legislators ignored this evidence?
Before Britain’s seat belt law was passed there had been eight debates in Parliament about it over the previous ten years. The main debate that resulted in the passing of Britain’s law was held on 28 July 1981. In this debate, a research report of mine, published earlier in the year, was much discussed, and much derided. It had a few eloquent libertarian supporters who liked its conclusions, but they were outnumbered by paternalistic health and safety enthusiasts who didn’t. Most of the parliamentarians queuing up to praise me or denounce me appeared not to have read my report – only my conclusions, which they liked or disliked. All of those who praised me, and my denouncers, were already established opponents or supporters of a seat belt law – no minds were changed by my evidence.
However within the Department of Transport, the promoters of the seat belt bill, my study had raised concerns. The Department commissioned a critique of my report by J E Isles. His report examined evidence from eight European countries (a subset of the 18 examined in my report) that had passed seat belt laws. He concluded that a law making the wearing of seat belts compulsory “has not led to a detectable change in road death rates”. For promoters of the bill this was an inconvenient truth. The Isles report was dated April 1981, more than three months before the parliamentary debate that led to the passage of the legislation. But it was suppressed. It was not published, and was not allowed to inform that debate. The Isles Report did not see the light of day until its existence was disclosed by New Scientist in an article published on 7 February 1985 – more than three years too late.
In the 1981 Parliamentary debate opponents of the law described my report variously as “bogus”, “riddled with inaccuracies”, “eccentric”, “preposterous”, “spurious”, and “wrong”. One supporter of the law (Austin Mitchell MP) described my report as “the only one that the hon. and learned Gentleman [Ivan Lawrence MP] can dredge up.” The Secretary of State for Transport in his contribution to the debate described my risk compensation hypothesis as “dubious and not proven”, but made no mention of his own department’s study whose conclusions supported mine. And my principal champion (Ivan Lawrence) described my findings as “astonishing and unexpected”. Such, at the time was the response to explanations of road accident statistics that invoked the risk compensation hypothesis.
A year later, too late for the parliamentary debate, I was invited to present my report to a meeting of the Society of Automotive Engineers in Detroit. It subsequently achieved peer-review status and was published as an SAE publication. To date, 25 years later, I am aware of no critique that has refuted its evidence, or conclusion – no country that has passed a seat belt law can demonstrate that it has saved lives. And “risk compensation” – is now widely accepted, and at the time of posting this blog registered 93,000 Google hits.
Since interest in the subject has revived, at least in the United States, and for historians of the role of statistical skulduggery in the formation of policy, I have scanned my scribbled-over copy of the Isles Report and put it on my website as a PDF file.