Report identifies ‘major errors’ in Scotland’s safety camera stats
Transport Scotland is to change the way it reports casualty data at camera sites after an independent review highlighted “major errors” with the current approach (Local Transport Today).
The annual report, ‘Key Scottish safety camera programme statistics’, presents casualty data from more than 400 camera sites. The analysis compares three-year data from a baseline with a three-year period after cameras were installed. But the reports have made no allowance for trends in casualties or ‘regression to the mean’.
Following complaints from anti-camera campaigners including Idris Francis, Mike Maher of the Institute for Transport Studies at the University of Leeds, was commissioned to review the way the data is reported. His findings have prompted Transport Scotland to announce revisions to the way data is presented in the future, and this year’s bulletin has been delayed until October.
Jennifer McCahill, assistant statistician to the Scottish Safety Camera Programme, said: “The forthcoming bulletin will incorporate a section where a sample of sites will be selected and the effects of trend and regression to the mean will be allowed for when trying to estimate the effect that safety cameras have had on road traffic accidents.”
Professor Maher’’s report says that between 2001 and 2011 the number of fatal and serious collisions across Scotland fell by an average of 4% a year and the number of all injury collisions by 3%. He adds that “there seems no reason to doubt that the same trend would have occurred at camera sites…therefore any estimate of the effect of cameras should take account of trend”.
On regression to the mean, his report said: “When the decision to install a camera is based on high numbers of collisions in the preceding period (typically three years) and this same period is used to provide the ‘before’ data in a before-after comparison, there is the danger of regression to the mean.”
Maher recommends that Transport Scotland uses a statistical method known as Empirical Bayes to identify the regression to the mean effect. This method was used by researchers in the four-year evaluation report of safety cameras for the DfT in 2005.
Citing the 2005 report, Professor Maher says the researchers suggested that for fatal and serious collisions, the 54% overall reduction observed at camera sites could be attributed to 10% trend, 34% regression to the mean, and 10% to the cameras.
Idris Francis, an independent campaigner who runs a website called 'fight back with facts', said: "Transport Scotland's figures completely ignored trend, which over 10 years alone accounted for a 42% fall in Scotland KSI. The standard method is to measure KSI over three years, then install a camera and measure KSI again over the next three years - and adjust for trend, which is easy, but also for regression to the mean, which is difficult. They ignored both.
"They also ignored, as many previous analyses have done, that virtually all of the observed falls happened in the gap year between site selection and camera installation, which is absolutely critical."