The accident rates still apply per vehicle million km but as you say the non camera routes are in the minority so to be fair the study should have done more non camera routes to even things out. I'm not a statistician though so couldn't say how much such a small sample could affect the results. There is also no way of knowing which way it would have affected things as a larger sample might have shown an even bigger reduction for non camera routes but we can't know.
The original study was supposed to select motorway lengths with and without roadworks and with and without cameras. Without knowing which lengths were used and when again we can't say for certain they were comparing like with like. I'd hope they were as otherwise it negates the whole purpose of the study!
|