13 deaths in January, heading for worst in over 10 yrs

[Flinders]

I've been out less this month than normally in January. I went out today on foot and nearly came a cropper on the small rural road I walked down, it was so icy; I wrenched my shoulder just trying to save myself from one slip. I doubt very much that this increase is due to more people going out on bikes.

I just hope that we don't have any more. But I always hope that.

[TonyR]

Safety in number means that more cyclists results in proportionally fewer deaths. That is, if the number of cyclists goes up by 50%, maybe the number of deaths only goes up 40%.

18% actually. The number of deaths is proportional to the numbers cycling raised to the power 0.4 and the risk per individual goes as the numbers to the power -0.6 So a 50% increase in cycling leads to a 18% increase in deaths but a 22% fall in the individual risk of being killed.

[Tonyf33]

And another..cycling in numbers really isn't working is it, certainly not in London
http://www.standard.co.uk/news/london/c ... 89385.html

[TonyR]

And another..cycling in numbers really isn't working is it, certainly not in London
http://www.standard.co.uk/news/london/c ... 89385.html

It very much is working in London with cycling rates rising dramatically to where it is now the dominant form of transport in many parts of central London. And yet the ksi rate has risen very little if at all and what deaths there are are dominated by lorries.

http://www.standard.co.uk/news/london/m ... 71069.html

[iviehoff]

It is very unlikely that anything can be learned from looking at the figures for one month. When the average number of deaths per month is (say) 10, over a 10 year period you can reasonably expect to see individual months varying from 0 to the low 20s, just as purely random variation, without stretching the bounds of mathematical credulity.

It seems to be counter-intuitive to most people that so much variation can be just random; people just aren't hard-wired naturally to understand probabilities and statistics. There was some uber-death-week in London a year or 2 ago that seemed outrageously unlikely to most people. But a quick calculation showed it to be something that random fluctuation could bring up once every 20 or 30 years.

[TonyR]

There was some uber-death-week in London a year or 2 ago that seemed outrageously unlikely to most people. But a quick calculation showed it to be something that random fluctuation could bring up once every 20 or 30 years.

I'll repost my earlier link that explores that event in detail - http://onlinelibrary.wiley.com/doi/10.1 ... 0715.x/pdf

[iviehoff]

There was some uber-death-week in London a year or 2 ago that seemed outrageously unlikely to most people. But a quick calculation showed it to be something that random fluctuation could bring up once every 20 or 30 years.

I'll repost my earlier link that explores that event in detail - http://onlinelibrary.wiley.com/doi/10.1 ... 0715.x/pdf

I overlooked your earlier reference to that paper, thank you for drawing it to my attention.

So you come up with 2.5% chance of occurring in 40 years, ie, roughly once in 1600 years. Nevertheless, things that happen 1 in 1600 years do often turn up around about once in 1600 years, and if that happens to occur in your purview, you will see it and be surprised. There are so many once-in-a-millennium or two things that can come your notice, that many will occur during our lifetime, it's merely a question of deciding what is significant to be bothered with. It's like the statistician's joke, "I couldn't help noticing your telephone number, 3955 8291, that's just so amazing. What was the chance that your telephone number would be precisely 3955 8291, it's like 1 in a 100 million. That's just so cool." Of course the unlikely things that come to our attention are usually rather more significant than that, but it drives home the point that there are a very large potential number of coincidences waiting to happen, that it is hardly surprising so many of them do.

Whether there was or wasn't some curious underlying factor that caused that cluster of accidents, or purely outrageous chance, what it absolutely isn't is evidence of an increase in the underlying rate of accidents. We look at the accidents over a longer period and quickly find that there is no disturbance to the average rate of accidents over a longer period.

You would expect road accident deaths to be somewhat more clustered than a Poisson distribution would suggest, because of the occurrence of multiple death incidents, but this is a minor issue I relation to London cycling deaths since riders are predominantly solo in London. Indeed the fact that you found not a single fortnight with 4 or 5 deaths in an extended period tends to demonstrate that it is a factor that it is of barely any effect.

[TonyR]

So you come up with 2.5% chance of occurring in 40 years, ie, roughly once in 1600 years. Nevertheless, things that happen 1 in 1600 years do often turn up around about once in 1600 years, and if that happens to occur in your purview, you will see it and be surprised. There are so many once-in-a-millennium or two things that can come your notice, that many will occur during our lifetime, it's merely a question of deciding what is significant to be bothered with. It's like the statistician's joke, "I couldn't help noticing your telephone number, 3955 8291, that's just so amazing. What was the chance that your telephone number would be precisely 3955 8291, it's like 1 in a 100 million. That's just so cool." Of course the unlikely things that come to our attention are usually rather more significant than that, but it drives home the point that there are a very large potential number of coincidences waiting to happen, that it is hardly surprising so many of them do.

There is also the observational selection bias. If you have 1600 things with a 1 in 1600 years chance of happening then you will observe one of them happening every year on average. And when it happens you will observe the remarkable fact that something with a 1 in 1600 chance happened. What you don't observe is the other 1599 things with a similar chance that didn't happen.

[pwa]

I don't think these statistics tell us enough to draw any conclusions about trends. Weather could be playing a big part.

I also wonder whether the rise in numbers of cyclists in London and the other big conurbations means that urban cycling is a bigger part of the picture than it once was. Cycling in a city (at least in the UK) is more risky than cycling in the countryside, and it seems to me that this is the side of cycling that is growing fastest. I live in a rural area and my impression is that things have improved over the past decade. Car drivers are passing wider, and it is a long time since I last had any verbal abuse from another road user.

[AndyBSG]

Weather could be playing a big part.

This is something that confuses me in general.

There have already been a few vague murmerings that the spike of cycling deaths in recent weeks is becasue of the very mild january so more people are cycling and the spate of deaths last year also took place in unseasonably mild weather.

So, with that in mind why don't we see a comparable number of deaths during summer months?

[kwackers]

There have already been a few vague murmerings that the spike of cycling deaths in recent weeks is becasue of the very mild january so more people are cycling and the spate of deaths last year also took place in unseasonably mild weather.

More people cycling combined with poor road conditions and darkness?
(Just guessing here).

[pwa]

Yes, poor light conditions with the sun low in the sky, but mild weather leading to people riding in large numbers. That might help to produce a spike in the figures.

[AndyBSG]

And January gets off to a grim start as well...

http://road.cc/content/news/141903-tipper-truck-involved-another-london-cyclist-fatality

One thing that springs to mind, if a toy or a cooking item or power tool was causing those sort of percentages of deaths they would be banned

[pwa]

Trouble is, Andy, if you ban tipper trucks you more or less ban road repairs! We all benefit from the stuff that tipper trucks tip. It's just the way some of them are driven that needs to be modified.

[AndyBSG]

In all honesty I think half the problem with tipper trucks is that they are classed as construction vehicles not haulage so they are exempt from a lot of the safety features haulge lorries are such as side bars to knock people away rather than drag them under, extra mirrors, etc.

The fact their drivers are typically paid per delivery rather than an hourly wage is also a factor as it encourages them to drive recklessly to try and fit more drops in.