As of 20 June all the analysis has been done and results obtained – all that remains is to document the method and exlain how the reults prove conclusively that speed cameras achieve little or nothing
Over recent weeks I have resumed this work, but then been delayed by computer problems. They have now been resolved and I hope to be able to show meaningful results by the end of March or so.
Some while ago I analysed 4.7m casualty accidents in Britain in terms of their location within 1km squares, approximating to the area of a speed camera site. I then analysed only the 130,303 such sites that had suffered at least one 3 year period of 4 KSI injuries or more.
I attach here the most important of the 24 Excel sheets showing, for each police area and nationally, from left to right
Area.
Number of examples
Fatalities 3 years prior
Fatalities in the qualifying period
% fall.
And then the same sets of 3 figures for KSI, Serious, Slight and All injuries
(Please note that although the numbers are correct, a minor error in the programme that at the end of the process allocates names to police are numbers has resulted in Scotland being shown twice, and from memory one other mistake of that kind. That will be corrected asap.)
The % falls in each category of casualty are due in part to long term improving trends but also to regression to the mean, by which accident numbers tend to fall back to prior levels after the unusually large numbers in the qualifying period.
I am now part way through a new analysis, based on perhaps 6m accidents from 1985 to 2007, but this time adjusting the numbers for trend so that the reductions shown will be the average for those police areas and nationally due to regression to the mean alone.
This should prove to be by far the best method of estimating it yet devised. One result is however very clear from the first evaluation – that the % falls are always greatest for Fatal than Serious, and greater still than for Slight. This is an inevitable consequence of there being approximately 10 Serious and 50 Slight injuries for each fatality, so logically the chances of another of the same severity happening in the same place again must be lower.
The same factors result in the other obvious result, that the falls are much greater in more rural police areas i.e. less busy roads and fewer accidents.
In particular, in the most rural areas such as North Wales and Scotland, KSI for example routinely falls by 70% without a speed camera in sight.
This analysis – with which Professor Richard Allsop who wrote the RAC camera report in 2010 agrees, will provide solid evidence of the substantial significance of regression to the mean.