Cyclist killed in lorry collision

Page may contain affiliate links. Please see terms for details.

Pale Rider

Legendary Member
Oh dear, another female cyclist has died following a collision with a lorry.

There's no good news here, but it's worth pointing out the number of deaths in Greater London has declined dramatically.

I think this is the first death in the Greater London area for about 11 months, or it may be the second.

Either way, that is a big improvement over the dozen or so deaths per year that were happening.

http://www.bbc.co.uk/news/uk-england-london-36363106
 

ianrauk

Tattooed Beat Messiah
Location
Rides Ti2
Sad. Poor lady.

Have cycled on St James's Road many a time. Looks like it was on the road bridge which is very narrow. You have to take primary always lest motor vehicles squeeze past.
 

numbnuts

Legendary Member
In her 20s so young and a great loss to her family RIP
 

Pete Owens

Well-Known Member
I wonder if the 11 month gap was random chance or a sign that things have changed. Anyone got time to do the maths?
You expect to see clusters and gaps in random sequences. Indeed, if a statistician were shown a sequence where supposedly random events were happening regularly (eg a sequence of monthly "cycle deaths in London" figures where there was exactly one death within each calender month) they would be suspicious). So 11 months without a death is no more evidence of increased safety than 11 deaths in one month was evidence of asudden increase in danger a couple of years ago.
 

srw

It's a bit more complicated than that...
You expect to see clusters and gaps in random sequences. Indeed, if a statistician were shown a sequence where supposedly random events were happening regularly (eg a sequence of monthly "cycle deaths in London" figures where there was exactly one death within each calender month) they would be suspicious). So 11 months without a death is no more evidence of increased safety than 11 deaths in one month was evidence of asudden increase in danger a couple of years ago.
That depends on an awful lot of things, and it's rather more complex than you state.

The most appropriate basic model for cyclist deaths is probably the Poission process. Based on the incidence over the last few years it's a process with a mean of 11 or so (per year). The waiting time distribution for the Poisson process is the exponential distribution with the same single parameter as the Poission process. So if the model I've chosen is correct then the observation "waiting 11/12 years" is indeed extreme, with p-value e^-(11*11/12) ~= 0.004%.

All of which is nice (and I acknoweldge that the maths may well be wrong - it's been 20 years since I've needed to be able to calculate the sort of thing in detail), but not very helpful. The cluster of deaths you mention was about as unlikely.

The most likely thing is that the model I've chosen is wrong - because the number of deaths in London has been drifting gradually downwards while the amount of cycling has been rocketing upwards. And because when you compare the number of deaths and serious injuries cycling in London with the number of cycling journeys taken - or miles ridden - or hours ridden - you need an electron microscope to see them.
 

slowmotion

Quite dreadful
Location
lost somewhere
I'll probably be shot down in flames, but the statistics for cyclists killed, or KSI (I forget which) are not nearly as bad as some non-cyclists believe. I can't remember if it's nationally, or London based but per mile ridden on a bike and driven by car, the ratio is 110 to 70. Both too high of course, but quite heartening to me.
 
Top Bottom