The paper is doing some weird things. To quote from it:
Fatality rates are difficult to obtain as there is no estimate of the miles cycled per year. We have compromised by providing the fatality rates per 100,000 estimated cyclists per kilometre on the roads of London. However, for an individual cyclist, it would be more useful to know the fatality rate per million miles cycled—a figure that currently cannot be obtained.
No, per 100k cyclists per kilometre; that is, per 100k cyclist-kilometres, which is dimensionally correct. Or am I misunderstanding what it is you find weird?
I only glanced at the paper, but my suspicion is that they are using something like 100K cyclists per kilometre of road, not per kilometre actually cycled. They admit to not knowing the miles cycled, but if you guesstimate the number of cyclists and you know the length of roads in London, you can produce a “cyclists per kilometre of road” metric. I am not sure how meaningful it is.
Cyclists per kilometre of road is a measure of cyclist density. To do a proper density estimate you also need to know how long (in hours) does an average cyclist spend on the road, but conceivably you can handwave it away as a near-constant. Given this, their approach is to look at the fatalities as a function of the cyclist density.
They are still missing too many variables to produce a useful estimate, but it’s not prima facie insane.
Yes it is prima facie insane, because the number of fatalities should look like f(density) times quantity of road (or quantity of cycling or something). Maybe they kinda get away with it when comparing fatalities across years, because the quantity of road doesn’t change much—but that would equally justify measuring quantity of cycling as, say, “number of cyclists divided by latitude” or “number of cyclist-kilometres times number of letters in city name”. So, good try, but I still think it’s obviously bonkers.
A foolish consistency is the hobgoblin of PEOPLE WHO NEED THEIR HEADS EXAMINED—GJMerson.
[EDITED to add:] Anyway, unless I’m misunderstanding your “Tony Stark” comment I don’t get the impression that you really find their use of that statistic very defensible.
Should I argue with this or no? Ah, a nice Catch-22… :-)
However there wasn’t much arguing in this thread. I didn’t tell gjm that he was wrong. Instead I offered more of an alternate explanation (to the “prima facie insane” hypothesis :-D) as I am generally interested in finding alternate ways of looking at things.
Once someone was saying that “you always say I’m wrong,” and I said, “I said you were right on occasions A, B, and C,” and they responded, “See! you’re doing it again! that proves you always say I’m wrong!”
Deaths per cyclist per kilometre of road is a crazy unit of measurement. I mean, sometimes you have to report the statistic you’ve got rather than the statistic you’d like, but I don’t see what possible practical significance this has.
The statistic we’d like to know is deaths per kilometre cycled. The average person in the UK cycles about 60 km a year (source: Department for Transport) and the population of London is about 8.5 million (source: Wikipedia), so the 19 deaths in 2006 correspond to about 3.5 deaths per 100 million kilometres cycled.
This is slightly higher than the UK average of 3.1 deaths per 100 million kilometres cycled, and on the high side for Western Europe (compare Netherlands: 1.0; Germany: 1.8; France: 3.1; Italy: 3.4).
The paper is doing some weird things. To quote from it:
Note: cyclists per kilometre.
No, per 100k cyclists per kilometre; that is, per 100k cyclist-kilometres, which is dimensionally correct. Or am I misunderstanding what it is you find weird?
I only glanced at the paper, but my suspicion is that they are using something like 100K cyclists per kilometre of road, not per kilometre actually cycled. They admit to not knowing the miles cycled, but if you guesstimate the number of cyclists and you know the length of roads in London, you can produce a “cyclists per kilometre of road” metric. I am not sure how meaningful it is.
Bloody hell, you appear to be right. What a ridiculous figure to be publishing.
I am quite sure how meaningful it is: not meaningful at all.
Paging Tony Stark! :-)
Cyclists per kilometre of road is a measure of cyclist density. To do a proper density estimate you also need to know how long (in hours) does an average cyclist spend on the road, but conceivably you can handwave it away as a near-constant. Given this, their approach is to look at the fatalities as a function of the cyclist density.
They are still missing too many variables to produce a useful estimate, but it’s not prima facie insane.
Yes it is prima facie insane, because the number of fatalities should look like f(density) times quantity of road (or quantity of cycling or something). Maybe they kinda get away with it when comparing fatalities across years, because the quantity of road doesn’t change much—but that would equally justify measuring quantity of cycling as, say, “number of cyclists divided by latitude” or “number of cyclist-kilometres times number of letters in city name”. So, good try, but I still think it’s obviously bonkers.
Funny how the usual roles reversed: I get charitable and you go THESE PEOPLE NEED THEIR HEAD EXAMINED :-D
A foolish consistency is the hobgoblin of PEOPLE WHO NEED THEIR HEADS EXAMINED—GJMerson.
[EDITED to add:] Anyway, unless I’m misunderstanding your “Tony Stark” comment I don’t get the impression that you really find their use of that statistic very defensible.
The actual explanation for this is your desire to argue with your current interlocutor, whoever that may be at the moment.
Should I argue with this or no? Ah, a nice Catch-22… :-)
However there wasn’t much arguing in this thread. I didn’t tell gjm that he was wrong. Instead I offered more of an alternate explanation (to the “prima facie insane” hypothesis :-D) as I am generally interested in finding alternate ways of looking at things.
Once someone was saying that “you always say I’m wrong,” and I said, “I said you were right on occasions A, B, and C,” and they responded, “See! you’re doing it again! that proves you always say I’m wrong!”
Deaths per cyclist per kilometre of road is a crazy unit of measurement. I mean, sometimes you have to report the statistic you’ve got rather than the statistic you’d like, but I don’t see what possible practical significance this has.
The statistic we’d like to know is deaths per kilometre cycled. The average person in the UK cycles about 60 km a year (source: Department for Transport) and the population of London is about 8.5 million (source: Wikipedia), so the 19 deaths in 2006 correspond to about 3.5 deaths per 100 million kilometres cycled.
This is slightly higher than the UK average of 3.1 deaths per 100 million kilometres cycled, and on the high side for Western Europe (compare Netherlands: 1.0; Germany: 1.8; France: 3.1; Italy: 3.4).