Helen
I am sure that you will be directed to Paul Smith's estimates. He has done the best job he can from the data he has sourced.
But it is a very difficult task because there are international influences on road trauma (car design and safety improvements are in general now dtermined internationally) as well as country wide and local influences.
For example since 1975 all countries recognised as reducing road trauma per 100 million kilometres have achieved drops in the range of 69% - 78% - yet the means they have achieved those gains vary widely.
THe standout exception to this trend is the USA which has only achieved a drop of 57%. Would be interesting to determine why that is.
In my own country Australia, all states and territories have achieved gains in the range above except one. Victoria has taken a horrendous speed enforcement approach since 1990 (budgetted speeding fine revenue for this financial year of A$300 - $400 million for 3.1 million drivers). Other states have taken a much less intensive speed enforcment approach. Yet similar results.
The exception is the Northern Territory with a 61% drop. It still has an unlimited rural speed limit, has by far the largest percentage of indigenous people (27%) has a low population density so most roads (except urban and two major highways) are unsealed, and has a more "gung ho" entrepreneurial appraoch to life on average than most other states (Its fatality rates per 100 million km are about 3-4 times higher than other states).
Finally in regard to individual studies at sites where cameras have been installed, I am not aware of any UK study that has been done in a way that would have eliminated regression to the mean effects. That requires for instance that you would go through the process of selecting candidate sites for speed cameras, and sort them into groups/ types based on site characteristics, severity of the crash problem et cetera. You would then as far as possible pair them off and randomly install a speed camera at one of the pairs but not at the other.
You would then analyse crash rates reductions for the two groups and in theory the difference would be due to speed cameras.
Note that regression to the mean can result in very large drops in crashes with no treatment depending on the black spot selection criteria.
Pauls estimate of the INCREASE in deaths due to the emphasis on speed cameras and speed enforcement are shown at
http://www.safespeed.org.uk/6800.xls His analysis suggests that the speed camera obsession has resulted in the fatality rate being 20% higher than it would have been if efforts had gone continuing with the sort of road safety initiatives that were used up till 1994. Included in these were in effect a higher police presence on roads.
I have questions about Paul's analytical approach, but recommend it to you so that you are open to the possibility that the obsession with speed cameras and speed limits has been counterproductive.
Regards
John Lambert
A road safety researcher in Australia (though not part of the mainstream of speed limit enforcement and speed camera zealots)
While I do not