This file below explains how regression to the mean, the natural tendency of most changing parameters to revert to their long term trends after a disturbance, in fact accounts for most or all of the observed falls in casualties after a speed camera – or for that matter a garden gnome – is installed at the side of the road.
My analysis of 3.8m accidents is now complete and explains that while it it mathematically impossible accurately accurately to quantify the effect of RTM using available datam it is easy to eliminate its effects entirely by analysing only post-installation data.