P
Paul Smith
Guest
On 20 Feb 2003 14:27:22 -0800, [email protected] (Richard Waters) wrote:
>> Further to discussions yesterday I have made a draft page and spreadsheet available concerning
>> the UK's loss of previous fatal accident trend in 1993.
>> http://www.safespeed.org.uk/fatality.html
>> Comments welcome. (not cross posted)
>Your mathematics may be correct (although I dispute the validity of picking arbitrary trend lines
>based on the overall change between two arbitrary years; you have not demonstrated that there is a
>statistically significant change since 1993; the data points are well-distributed and the chances
>of seeing a 'trend' such as the one you identify purely by blind chance are all too present. The
>onus is on you to show that the trend is not attributable to lady luck if you wish to use these
>figures to support a case for change in public policy).
I'll do whatever I can to build the case.
>Any such analysis should anticipate a gradual fall in the year-on-year percentage change since
>the eventual outcome of any road safety policy will be to leave us with a residual level of
>accidents that reflect human and technical inadequacy (we will never achieve an accident-free
>road network while humans drive, walk and cycle). Each new development in road safety may
>reasonably be expected to cause a temporary spike in this for a few years as their impact is felt
>and then, over subsequent years, the rate of change will fall away as there is no new advance to
>reduce the figures.
Exponential trends already have such a characteristic. The trends identified are exponential.
>Even if I grant that that the maths is acceptable, the conclusions you draw from it hold no water.
Naturally I disagree.
>Your assertion that speed cameras and the speed kills policy are the only possible cause is at
>worst ridiculous and at best an assertion unsupported by any causal evidence. The statement that it
>must be true because it is the only reasonable cause is infantile, and your attempt to hide such a
>random leap of illogic within a mass of reasonable-sounding mathematice akin to the defenses of
>Astrology that are littered astronomical terms to impress the gullible.
There's no way to prove the case at present. But evidence is building everywhere I look.
>So let me offer you one alternative hypothesis. In 1992 the Ford Escort was first fitted with an
>airbag. This happened on a few K reg models (available Autumn 1993) and became more widespread over
>subsequent years (http://www.parkers.co.uk/pricing/used_options/usedoptions.asp?model_id=389).
>Airbags became widely available in the Vauxhall Astra in 1993
>(http://www.parkers.co.uk/pricing/used_options/usedoptions.asp?model_id=234). These 2 cars account
>for a large percentage of UK sales. This is not the first introduction of airbags into UK cars, but
>it is when they became common.
>1993 as a year saw a large percentage reduction in the percentage change of fatalities per billion
>vehicle miles (roughly 1% compared to .6% in 1991). This reflects the impact of airbags in reducing
>the fatalities of drivers. Since then the percentage change of deaths per billion KM has been
>dropping away steadily. I put this down to drivers adapting to the presence of airbags, feeling
>safer as a result and driving less safely (there are figures below to support this).
If such a risk compensation effect was happening now with airbags, you would have to explain why it
didn't happen in the past with disc brakes, radial ply tyres, seatbelts, crash testing, side impact
protection, head rests, etc.
>Mr. Smith, you are selling snake oil. Smoke and mirrors, smoke and mirrors.
You are entirely mistaken. But thanks for your comments.
> - Richard Waters
>
>Year Bill. deaths death rate percentage change in death vehicle per vehicle rate per mile miles
>mile 1992 412.1 4,229 10.2615494 -0.837332209 1993 412.2 3,814 9.252610333 -1.008939067 1994 422.6
>3,650 8.637008992 -0.615601341 1995 430.9 3,621 8.404082978 -0.232926014 1996 442.5 3,598
>8.131073446 -0.273009532 1997 452.5 3,599 7.95359116 -0.177482286 1998 459.2 3,421 7.449912892
>-0.503678268 1999 466.0 3,423 7.345272889 -0.104640003 2000 467.7 3,409 7.288860381 -0.056412508
>2001 473.7 3,450 7.283090564 -0.005769817
Notice how 1998 was the only year since 1994 to fail to show a worse result than the previous year?
There's nothing else like this in the history of GB road fatality rates.
--
Paul Smith Scotland, UK http://www.safespeed.org.uk please remove "XYZ" to reply by email Let's make
speed cameras as unacceptable as drink driving
>> Further to discussions yesterday I have made a draft page and spreadsheet available concerning
>> the UK's loss of previous fatal accident trend in 1993.
>> http://www.safespeed.org.uk/fatality.html
>> Comments welcome. (not cross posted)
>Your mathematics may be correct (although I dispute the validity of picking arbitrary trend lines
>based on the overall change between two arbitrary years; you have not demonstrated that there is a
>statistically significant change since 1993; the data points are well-distributed and the chances
>of seeing a 'trend' such as the one you identify purely by blind chance are all too present. The
>onus is on you to show that the trend is not attributable to lady luck if you wish to use these
>figures to support a case for change in public policy).
I'll do whatever I can to build the case.
>Any such analysis should anticipate a gradual fall in the year-on-year percentage change since
>the eventual outcome of any road safety policy will be to leave us with a residual level of
>accidents that reflect human and technical inadequacy (we will never achieve an accident-free
>road network while humans drive, walk and cycle). Each new development in road safety may
>reasonably be expected to cause a temporary spike in this for a few years as their impact is felt
>and then, over subsequent years, the rate of change will fall away as there is no new advance to
>reduce the figures.
Exponential trends already have such a characteristic. The trends identified are exponential.
>Even if I grant that that the maths is acceptable, the conclusions you draw from it hold no water.
Naturally I disagree.
>Your assertion that speed cameras and the speed kills policy are the only possible cause is at
>worst ridiculous and at best an assertion unsupported by any causal evidence. The statement that it
>must be true because it is the only reasonable cause is infantile, and your attempt to hide such a
>random leap of illogic within a mass of reasonable-sounding mathematice akin to the defenses of
>Astrology that are littered astronomical terms to impress the gullible.
There's no way to prove the case at present. But evidence is building everywhere I look.
>So let me offer you one alternative hypothesis. In 1992 the Ford Escort was first fitted with an
>airbag. This happened on a few K reg models (available Autumn 1993) and became more widespread over
>subsequent years (http://www.parkers.co.uk/pricing/used_options/usedoptions.asp?model_id=389).
>Airbags became widely available in the Vauxhall Astra in 1993
>(http://www.parkers.co.uk/pricing/used_options/usedoptions.asp?model_id=234). These 2 cars account
>for a large percentage of UK sales. This is not the first introduction of airbags into UK cars, but
>it is when they became common.
>1993 as a year saw a large percentage reduction in the percentage change of fatalities per billion
>vehicle miles (roughly 1% compared to .6% in 1991). This reflects the impact of airbags in reducing
>the fatalities of drivers. Since then the percentage change of deaths per billion KM has been
>dropping away steadily. I put this down to drivers adapting to the presence of airbags, feeling
>safer as a result and driving less safely (there are figures below to support this).
If such a risk compensation effect was happening now with airbags, you would have to explain why it
didn't happen in the past with disc brakes, radial ply tyres, seatbelts, crash testing, side impact
protection, head rests, etc.
>Mr. Smith, you are selling snake oil. Smoke and mirrors, smoke and mirrors.
You are entirely mistaken. But thanks for your comments.
> - Richard Waters
>
>Year Bill. deaths death rate percentage change in death vehicle per vehicle rate per mile miles
>mile 1992 412.1 4,229 10.2615494 -0.837332209 1993 412.2 3,814 9.252610333 -1.008939067 1994 422.6
>3,650 8.637008992 -0.615601341 1995 430.9 3,621 8.404082978 -0.232926014 1996 442.5 3,598
>8.131073446 -0.273009532 1997 452.5 3,599 7.95359116 -0.177482286 1998 459.2 3,421 7.449912892
>-0.503678268 1999 466.0 3,423 7.345272889 -0.104640003 2000 467.7 3,409 7.288860381 -0.056412508
>2001 473.7 3,450 7.283090564 -0.005769817
Notice how 1998 was the only year since 1994 to fail to show a worse result than the previous year?
There's nothing else like this in the history of GB road fatality rates.
--
Paul Smith Scotland, UK http://www.safespeed.org.uk please remove "XYZ" to reply by email Let's make
speed cameras as unacceptable as drink driving