In a recent post, I criticised British sociologist Alan Buckingham who gave a number of presentations for the CIS criticising speeding laws and their enforcement. I made the point that Buckingham was not presenting research-based arguments but a series of sloppy and misleading arguments(while I usually disagree with assumptions that guide CIS research, and the results that flow from those assumptions, the quality is generally pretty good, unlike some other rightwing thinktanks) . It emerged in the ensuing discussion that Buckingham wasn’t even presenting original arguments of his own but was restating and endorsing those of a British lobby group, Safespeed, of which he’s a member. In a statement on its website , Safespeed said

Dr Buckingham is a Safe Speed member and frequent visitor. It is obvious that his article draws heavily on our material, but unfortunately the print version of the article appeared without giving Safe Speed the required credit. We’re happy to accept that this was an oversight and not a deliberate attempt to infringe our copyright. The PDF version of the article has been officially amended to include the necessary acknowledgement. (Thanks to Don Arthur for this link. On checking, the statement is not there any more. It has apparently been withdrawn while a new one is being drafted. If anyone has a cached version, I’d appreciate it).

Having criticised Buckingham for not doing, or using, the right kind of research, I should say something about what kind of research is appropriate. Although I’ve done research on the economic aspects of road crashes and law enforcement, I’m not an expert on the effectiveness of particular road safety measures. However, reader Mark Leggett has kindly supplied me with this study by Newstead, Cameron and LeggettFor those who don’t want to read the whole thing, here’s the abstract

Random Road Watch (RRW) is a traffic policing program in operation in Queensland, Australia. It differs from conventional traffic policing in that an explicit resource management technique is used which randomly schedules low levels of police enforcement in a manner intended to provide long-term, widespread coverage of a road network and hence maximise road safety benefits. Implementation of the program studied in Queensland covered 55% of total crashes within the state. This study aimed to measure the crash effects of the RRW program in Queensland. A quasi-experimental study design was used for the evaluation incorporating Poisson regression statistical analysis techniques. Analysis of the effects of the Queensland RRW program on crash frequency has shown the program to be effective overall. Estimated program effects were largest on fatal crashes, with an estimated reduction of 31%. Estimated aggregate program crash effects reduced with crash severity and increased with time after program introduction. Crash reductions in the third year after program introduction translated into savings, at state level, of some
12% of the state’s crashes of all severities and some 15% of the state’s fatal road crashes. Overall, the program produced a significant 11% reduction in total crashes in areas outside of metropolitan Brisbane. The opportunity-cost benefit:cost ratio for the program was estimated to be 55:1.

I should say that I don’t claim that this study is conclusive or that its results are necessarily applicable to other road safety initiatives. In addition, I would have preferred a broader benefit:cost analysis, though from the data here it’s clear that the result would have been the same. But this is the kind of research that’s needed if you want to make claims about the effectiveness of specific road safety measures, such as speed cameras.

The time series evidence of dramatically declining death rates, and the comparison with rising death rates in the US, provides pretty conclusive evidence of the effectiveness of road safety measures in general. But to look at individual measures you need detailed research of the kind undertaken by Newstead et al. To try, as Buckingham does, to link second-derivatives of the death rate (a slowdown in the rate of decline) to specific initiatives such as speed cameras, is silly, and would not stand up to the kind of statistical scrutiny that is required in serious academic work.