An ABD analysis of accident figures in these six forces shows that to ensure the hypothecation "trials" produced the desired result for the Home Office, the odds were deliberately skewed by carefully selection of the trial areas.
The black lines on these graphs show actual fatal road accidents in each of the six areas area over the period 1993 to 1999.
The coloured band shows the statistically expected minimum and maximum fatalities during the period.
Assumes annual toll normally distributed about mean, with 95% confidence limits; 2 standard deviations either side of the mean.
|Fatal accidents||Statistically Expected Maximum||Statistically Expected Minimum|
The general trend is downwards, but notice the relatively small numbers due to Cleveland being a small area, these will have little effect on the overall result.
A very significant increase in fatal accidents occurred in 1999.
It exceeds the expected maximum, so is almost certain to be lower in 2000.
Another significant increase, virtually guaranteeing a decrease in 2000.
Well what a surprise.
How will they explain this one?
A generally increasing trend, but note the high numbers due to the size of the force area.
Now let's put all six of these areas together and see what we get:
All six areas
Hello, hello, hello, what's this then? Fatal accidents at the statistical maximum in 1999, meaning they are almost certain to be lower in 2000.
Notice how the blip in 1995 is caused almost exclusively by a blip in the Nottinghamshire figures. The slight drop in accidents between 1998 to 1999 in Nottinghamshire and Thames Valley, is completely masked by the rises in the other four areas.
Compare this with the figures for the whole of England:
All of England
So, the six areas were specifically chosen because the Home Office knew that statistically, the fatal accident figures for the six regions combined in 2000 were almost certain to be lower than 1999. This effect is kown as 'regression to the mean'.
'Regression towards the mean' is like the old saying 'lighting never strikes twice'. Infact lighting may strike twice at the same spot, but as lighting is largely random in nature, if it strikes once, it is more likely to strike somewhere else next time. Thus it is likely not to strike at the same spot.
Accidents are also random events. So if an unusually large number occur in one year, then it is highly likely (and a good bet for the government) that accidents will reduce the next year.
Imagine a football team win by 10-1. Do you think a bookmaker would give you good odds on them scoring fewer goals in their next match?
This fabricated 'evidence' would then be used to 'prove' that speed cameras were a success in saving lives, and would be cited as justification for rolling out the speed camera hypothecation project to the whole country.
The Six v England
On Monday 13th August 2001, the Home Office announced that the speed camera trial had been — wait for it — wait for it — a success! — Accidents had reduced between 1999 and 2000!
Well isn't that amazing children?
Who would have guessed it?
Once again the government is attempting to hoodwink the public by selecting statistics in isolation, telling how accidents fell between 1999 and 2000, without telling what happened before 1999.
This is exactly the same trick they used to claim that the first speed cameras in the country were a success -
see our "Great Speed Camera Lie of London" page.
Help us stop the lies!
The official government report from the DTLR
(Note their usual claim that a 1 mph reduction in speed equates to a 5% reduction in accidents, see our demolition of this patently absurd claim here)