On 11/10/17 12:28, Peter Parry wrote:
On Wed, 11 Oct 2017 10:20:28 +0100, TMS320 wrote:

On 11/10/17 09:35, Peter Parry wrote:
On Tue, 10 Oct 2017 12:06:02 -0700 (PDT), Simon Jester

Do these figure take into account that the majority of car miles
are on trunk roads where there are few, if any, pedestrians? Whilst
bicycles spend most of their time in urban environments.

If not then they are worthless.

They do. On urban roads Pushbikes seriously injure 26 pedestrians
per billion km and kill 0.5, cars seriously injure 10 and kill 0.7.

So you love to push this.

"Push"? One mention of DfT statistics is pushing them?

Yes. You have brought this up several times before. It is obvious you
think it is meaningful.

Irrespective of car bike/figures, explain why
you think pedestrian casualties per vehicle-distance is meaningful.

What other measure would you prefer? Per journey is pretty
meaningless and almost impossible to measure.

I asked you why you think pedestrian casualties per vehicle-distance is
meaningful. Don't evade the question.

Per trip is not perfect but less dependant on population distribution.

An ongoing problem with pushbike accident statistics is their inherent
inaccuracy. It is widely accepted that they vastly underestimate the
number of accidents involving cyclists as most go unreported in
Stats19, the major source of accident data collected by police
attending accidents. Unlike motor vehicles where non-reported
accident data (on both insured and uninsured drivers) is available to
researchers from sources such as the Motor Insurers Bureau few
cyclists carry specific insurance and there is no equivalent of the
MIB to compensate people injured by cyclists so the number of
non-reported accidents is impossible to determine with any accuracy.

Distances travelled are also very inaccurate. Questions in the census
are poorly framed and in the last one I couldn't put myself down as a
bicycle user. With motor vehicles it is easier to correlate distance
against things such as fuel usage.

The one figure which can be relied upon is deaths - but the absolute
number of these is small so identifying trends or the effect of
measures to reduce them is difficult. Lumping in seriously injured
gives a bigger data set, but an inaccurate one. In many cycle related
accidents, especially those where no one else was involved, the event
does not appear in Stats19 as the police are not called. Go beyond
that to "minor" accidents where injury nonetheless occurs and the data
on cycling accidents is simply not collected in any reliable way at
all.

An SVA is not relevant to pedestrian casualties. I have little concern
about self inflicted injuries.

Try
to bear in mind my earlier point about figures from France that show no
connection.

Perhaps I have misunderstood your claim but it seem implausible.
French figures for 2010-2012 show wide variations in accident death
rates per million inhabitants in 95 districts ranging from 17 in
Hauts-de-Seine to 142 in Lozère. That shows all, not just cycling
related accidents but it seems unlikely that there will be a vast
difference in rates for push bikes. - do you have a link to the source
document?

Only pedestrian deaths are being discussed here. In 2001 in Lozerre
there were 9 deaths (122 per million), zero in 2002, 1 in 2003, 1 in
2005: in Haute-de-Seine there were 15 in 2001 and 12 in 2002 (8.4 per
million). In 2002 the average was 14.55 per million with only 18
departments above 20. It looks like something exceptional happened in
Lozerre in 2001.

The data was taken from data.gouv.fr. The data for the years I have is
no longer available and in recent years they have reduced the breakdown.