Dangerous Drivers

[binary_star]

Loz Wrote:

-------------------------------------------------------

> LadyDeliah Wrote:

> --------------------------------------------------

> -----

> > I agree the drafting was imprecise, but it

> clearly says that use your own parameters to prove

> cycling

> > (general) is more dangerous. It does not say,

> use your own paraneters to prove cycling is more

> dangerous

> > within those parameters.

>

> Actually, the exact wording was "I've asked anyone

> who asserts any different to pick any parameter

> for danger they like". I chose relative deaths of

> pedestrians per mile travelled across all road

> types.

>

> QED, I believe. Challenge aced.

We both know those stats do not indicate cyclists are more of a danger to pedestrians. It's your 'across all road types' that's the problem isn't it? Just because cars can travel thousands and thousands of miles without a pedestrian in sight and not hit one doesn't really mean they are less dangerous does it? It means you've chosen the wrong way to measure danger. You're arguing that you're right on a technicality because I let you choose anything to measure danger. So 'pick any parameter for danger' was badly worded....you could have picked 'number of hedgehogs decapitated' or some other such nonsense so I have to admit you got me on that one.

Loz Wrote:

-------------------------------------------------------

> LadyDeliah Wrote:

> --------------------------------------------------

> -----

> > I agree the drafting was imprecise, but it

> clearly says that use your own parameters to prove

> cycling

> > (general) is more dangerous. It does not say,

> use your own paraneters to prove cycling is more

> > dangerous within those parameters. Otherwise you

> could rightly say, parked card cause zero deaths

> > per mile compared to moving cyclists per mile.

>

> But that is exactly how stats work.

No it's not. LadyD was right I'm afraid. You can have a statistically significant result but still draw the wrong conclusion from it (which is what you have done here). For instance I could conclude that cancer causes smoking because smoking and cancer are highly correlated. You have introduced bias because you're not measuring the right variable (miles and miles of unpedestrianised roads). You've already introduced bias before you've begun to analyse the data.

Maybe I should have been explicit: "pick any parameter that measures danger" perhaps?

[binary_star]

Loz Wrote:

-------------------------------------------------------

> But actually, it was a pretty interesting

> conclusion anyway, and one I bet you didn't expect

> - even given the parameters. I suspect that, even

> if you do whittle it down, the best you could hope

> for is something close to parity, given the

> wideness of the 2:1 result I got originally.

I am genuinely interested to find out but I can't open the data tables on my phone. I would be surprised if mile per mile on minor roads cyclists kill more peds but it could be true (and I'd be disappointed if it were no doubt about it!) I take my hat off to you though for your creativity, although to really be right even on that technicality you'd need to change your parameter to 'deaths of pedestrians per ESTIMATED mile travelled across all road types"

[binary_star]

Loz Wrote:

-------------------------------------------------------

> That is my analysis.

But not mine, I have drawn and wanted to share an alternative conclusion.

Loz Wrote:

-------------------------------------------------------

> Just sitting there sniping and trying to pick holes is lazy debating.

See above.

Loz Wrote:

-------------------------------------------------------

But, to answer one point: "You can't use traffic mile figures from traffic miles that don't include cyclists..?". Of course it is fair, as long as there are pedestrian casualties - I thought that would be obvious?

Yep you're right, but there need to be pedestrians for there to be pedestrian casualties. What I should have said was:

"You can't use traffic mile figures from traffic miles that don't include pedestrians. "

I know you discounted motorways but there are many other roads with little or no pedestrians or cyclists (and not just because they're banned). The fact we're including them in the data means we're allowing motorists (but not cyclists) to clock up thousands and thousands of miles in places where there is little or no chance of seeing a pedestrian let alone hitting one.

We're told by your data that cyclists aren't really using these roads but motorists can clock up a large number of pedestrian free (and therefore pedestrian death free) miles v quickly (like a-roads/national speed limit country roads with no footpaths).

My analysis of the data would be that deaths per mile are lower for cars but we can't attribute that to them being less dangerous...it could simply be that they travel a lot more miles where there just aren't pedestrians to kill.

Loz Wrote:

-------------------------------------------------------

If you want to go into the mess of trying to separate different road types then back it with sourced, reputable stats.

Your source separated the road by type which I was hoping to use but I can't see the raw data.

I'm disappointed in your analysis because you've tried to use what you've identified as a flaw (which essentially amount to a loophole) in the original 'challenge' to try to win on a technicality, because the variable you've chosen to measure 'danger to pedestrians' doesn't really measure that at all.

M'Lord!

[binary_star]

Alright, I managed to open those data tables, so I'll take the bait. Normally I wouldn't bother but the constant 'cyclists are dangerous' mantra gets on my tits which was my point in the first place, so for Loz...

The relatively small number of pedestrian deaths by cyclists means even minor differences will massively skew the data year on year if you're going to measure deaths per mile. However, I've voiced my opinion on the problems with 'deaths per estimated mile' enough so let's run with it for now...

If you are going to compare two data sets you should try to pick the same year to do it (i.e. not compare my 2007 deaths with your 2010 estimated travel miles). The reason I used 2007 data for pedestrian deaths was because that year was a bit of an anomaly for cyclists in that they killed a record 6 pedestrians. If you look at most other years its actually more like 2 or 3, in fact in 2009 there were none, so that year cyclists weren't deadly at all to pedestrians using 'estimated death per mile' or any other parameter because they didn't kill any. If I use that year, I don't need to do any calculations...cyclists by your own definition were not dangerous at all.

I think that's a poor 'win' for my argument, so let's continue...

You were using 2010 data so I'll work out pedestrian deaths in 2010 (from the ONS stats you used here: https://docs.google.com/spreadsheet/lv?key=0AonYZs4MzlZbdEZBNS1ETG8xT0JBSnR5N3Z6Q0hzNnc&f=true&noheader=false&gid=13) per estimated billion vehicle miles travelled in 2010 (from the dft stats you used here: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/10424/tra0104.xls)

PEDESTRIAN DEATHS 2010

--------------------------------

By car or van: 133

By pedal cycle: 2

Estimated billion miles travelled on ALL ROADS, EXCLUDING MOTORWAYS 2010

---------------------------

Cars, taxis (194.1) + vans (33.6) = 227.7

Pedal cycles: 3.1

Results in number of deaths per billion mile

By cars + vans: 133/227.7 = 0.584

By pedal cycles: 2/3.1 = 0.645

Estimated billion miles travelled on

MINOR URBAN RDS 2010

---------------------------

Cars, taxis (51.8) + vans (8.7) = 60.5

Pedal cycles: 1.8

Results in number of deaths per billion mile

By cars + vans: 121/61.1 = 1.980

By pedal cycles: 2/1.8 = 1.111

If you look at the two calculations, it clearly shows that on roads where there are actually likely to be any pedestrians, cars are almost twice as dangerous. In general, things don't tend to pose a danger to you when they're nowhere near you... Even with your loophole allowance which allows cars to rack up almost 200billion miles worth of relatively pedestrian free roads (of which cyclists only used 0.6billion miles), cars are still not coming out much better than cyclists.

[Loz]

Q1: Where did you get the statistic for pedestrians killed by cars/vans on minor urban roads only?

[LadyDeliah]

I wondered why deaths from trucks, buses and coaches are not included in the stats. I don't think the debate should be restricted to cars v bicycles. All motorists should be included.

[Annette Curtain]

I wonder if there's stats for cyclists/pedestrians killed by cars with blacked out windows.

In fact, what's with the whole blacked out windows thing. I reserved my judgement until I drove a car with said 'blacked out windows' and fekk me YOU CANNOT SEE OUT OF THE DAMN THINGS

Sure I get SOME people don't want to be seen, but hey WE WANT YOU TO SEE US.

Really....

[LadyDeliah]

I think some nice net curtains would be more appropriate!

[LadyDeliah]

Little update.

Since purchasing my pretty enormous and unmissable helmet cam, I've noticed a significant improvement in behaviour of motorists who see it atop my bonce, with it's little red 'filming' light on.

It's been amazing.

I still get arseholes zooming past me from behind an inch or so from my handlebars if they haven't clocked the camera, but even those incidents are less frequent than before.

I may look like a tw@t with the camera on my head, but it's a fair trade-off :-)

[Otta]

Has anyone mooned the camera yet?

[LadyDeliah]

Lol, no. I haven't had that particular treat yet but I look forward to it with anticipation :-)

[binary_star]

Just out of interest, how is the helmet cam working out LD?

I bought one ages ago but when I chucked out my previous helmet the camera attachment was still on it so only got to use it a couple of times. Thinking of giving it another go though...

[LadyDeliah]

It's good. I have to clear the memory pretty much weekly, but I just save everything in the hope I'll get time to edit it some time in the future.