Accuracy of historical bicycle miles travelled and cyclist casualty rates.

User avatar
The utility cyclist
Posts: 1989
Joined: 22 Aug 2016, 12:28pm

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby The utility cyclist » 5 Jan 2019, 12:03am

thirdcrank wrote:I'd take some convincing that there's any benefit to cyclists or cycling from stats about casualty rates, although accuracy is always to be preferred.

The most obvious outcome is that air travel is somehow best and cycling the worst. All that tells is what we know already ie that cyclists are vulnerable in comparison with the occupants of motor vehicles. If it were to be possible to collect and compare data measuring distances ridden on, say, ordinary roads, cycle lanes, cycle tracks (ie segregated provision) that still wouldn't factor in the crap provision as compared with the decent stuff.

The case for cycling is already pretty obvious in terms of things like the health benefits and sustainability. It's also pretty obvious that motor vehicles can travel faster and so go further in a given time. The big problem is that policies which acknowledge this are politically unacceptable here.


Yes cyclists are vulnerable, however using (IHO) massively inaccurate distances travelled by people on cycles and then using that to come up with a figure of casualty rates per mile does in fact project an image that cycling is far more dangerous than it really is. This DOES have a negative effect on cycling in more than one respect.
One use of casualty rates/numbers is used time and again by many organisations to press for PPE use, it also makes for scary reading from an average joe/person who might be wanting to cycle on the roads but is put off because of the continual focus on persons hurt/killed who use a bike. The stats regarding cycle injuries/deaths and types of injuries are NEVER, ever compared to that of pedestrians, motorvehicle occupants and the general populous unless it's places like here, the fanfare re injury stats on the road are always fanfared when it comes to cycling so the rate of injuries is hugely important.

passing the junction
Posts: 6
Joined: 23 May 2018, 8:48pm

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby passing the junction » 6 Jan 2019, 2:46pm

I would say, in relation to the original question, that the annual road traffic estimates are good enough for year-on-year comparisons - at least I cannot think what would make them badly wrong. Having said that, as shown in the methodology note that's linked from TRA0401, the estimate is from:
1. a network of count points - the individual locations do change sometimes but I guess that is designed not make an overall difference nationally. But I don't know if at some point in the distant past, the overall roadtype selection changed to put more weight on roads where people are less likely to cycle or even vice versa.
2. converting a single 12-hour manual count into an 'Annualised Average Daily Flow'. This has resulted in some strange (ie wildly incorrect) cycling estimates for sites that I know of. But you can get the "raw" count figures as well as the AADF from the DfT site - looks as though it only goes back to the year 2000 though. I think there are some aspects of the methodology that probably make the cycle count less dependable,eg, counts can happen on designated 'neutral days' between March and October. Of course, in respect to numbers cycling, there can sometimes be a big difference between early March and mid-June.

Sure, the survey does not go out of it's way to find 'successful' cycle routes but in that way it is more representative - it just shows how far and few between success stories are in the national context.

Vorpal
Moderator
Posts: 16305
Joined: 19 Jan 2009, 3:34pm
Location: Not there ;)

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby Vorpal » 7 Jan 2019, 4:46pm

The other problem relates to the fact the cyclists can easily change mode, or leave the road and the counts only include cyclists on the main carriageway.

The DfT count used to be wildly inaccurate for the A120 near Marks Tey. I haven't checked it recently, but the counting point on the west side of the Marks Tey junction used to be located 30 or 40 feet on the wrong side of the point where most cyclists left the road. Cyclists approaching the junction would go off the A120 onto the drive / service road for the shops there, onto the pavement for a few feet, then down Station Road. They've widened the pavement there, and I imagine that means that cyclists go off the road earlier to avoid queuing traffic.

Folks who aren't going to the station are mostly either headed towards Aldham on the back lanes (so use Staiton Road), or to Marks Tey over the footbridge. The DfT count only got the 1 or 2 cyclists per day who went straight over the junction and down the B1408 towards Stanway.
“In some ways, it is easier to be a dissident, for then one is without responsibility.”
― Nelson Mandela, Long Walk to Freedom

User avatar
mjr
Posts: 12112
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby mjr » 7 Jan 2019, 5:29pm

passing the junction wrote:I would say, in relation to the original question, that the annual road traffic estimates are good enough for year-on-year comparisons - at least I cannot think what would make them badly wrong.

Really? The problem is that the manual counts are inaccurate in many ways like Vorpal describes, and then many get estimated most years based on the past inaccurate AADF and their forecasts.

For example, I just picked count point 8702 off the Norfolk list at random and it was only actually counted twice in the last ten years, in 2011 and 2016. All the other annual data are estimates.

That might be excusable if their forecasts were any good, but I attach a graphic illustrating DfT forecasting accuracy. I'm sure it won't come as any surprise that they consistently overestimate the increase in car use as pictured, but underestimate cycle use. This bad planning is helping to create a transport dystopia of pollution, congestion and ill health.
Image Attachments
bettertransport_dft_Forecasts_2015_social_sm.jpg
DfT forecasting :(
MJR, mostly pedalling 3-speed roadsters. KL+West Norfolk BUG incl social easy rides http://www.klwnbug.co.uk
All the above is CC-By-SA and no other implied copyright license to Cycle magazine.

thirdcrank
Posts: 28054
Joined: 9 Jan 2007, 2:44pm

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby thirdcrank » 7 Jan 2019, 5:37pm

The other part of the ratio is the number of casualties. Doubts have been expressed about the accuracy of recording. And I'm a bit surprised that if highwaymen and the like are thought to be making dodgy use of the stats, anybody thinks they will behave any differently whatever the numbers.

Vorpal
Moderator
Posts: 16305
Joined: 19 Jan 2009, 3:34pm
Location: Not there ;)

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby Vorpal » 8 Jan 2019, 1:17pm

Traffic forecast models are another thing altogether. The models that create forecasts are different to the models that estimate daily flow. completely different software is used.

There are far too many variables to easily forecast future traffic accurately:
-economic prosperity
-overall population
-population of drivers
-new development (housing, shops, schools, etc.)
-employment

Are all things that affect traffic development. One of the biggest problems with traffic forecast on a smaller scale is that an analyst in one contractor or local authority may not be aware of developments in neighboring areas that will affect traffic patterns, sometimes substantially.

In general, a good analyst can produce traffic forecasts that are reasonable on a systemic basis for traffic that makes up a proportion of the overall traffic greater than perhaps 2 or 3%. The reason that they include 'low' and 'high' estimates is account for the many variables over which they have no control. It's not that actual traffic cannot fall outside the estimates, but between high and low, they cover 75% of likely outcomes. Otherwise the range would be so big as to render the forecast useless.

Forecasts do not work well for small areas or specific locations, and they do not work well for any low volume modes of traffic. However, highways engineers often behave as if they do.
“In some ways, it is easier to be a dissident, for then one is without responsibility.”
― Nelson Mandela, Long Walk to Freedom

brynpoeth
Posts: 9045
Joined: 30 Nov 2013, 11:26am

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby brynpoeth » 8 Jan 2019, 1:40pm

No. New roads -> more [motron] traffic, Minus One :(
Cycling-of course, but it is far better on a Gillott
We love the three Es: enforcement, enforcement & enforcement

User avatar
mjr
Posts: 12112
Joined: 20 Jun 2011, 7:06pm
Location: Norfolk or Somerset, mostly
Contact:

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby mjr » 8 Jan 2019, 4:30pm

Vorpal wrote:Traffic forecast models are another thing altogether. The models that create forecasts are different to the models that estimate daily flow. completely different software is used.

Yes, but if the models that estimate daily flow are being fed into the models that create forecasts (because the actual traffic counts don't happen most years), then they both have effects.

Vorpal wrote:In general, a good analyst can produce traffic forecasts that are reasonable on a systemic basis for traffic that makes up a proportion of the overall traffic greater than perhaps 2 or 3%. The reason that they include 'low' and 'high' estimates is account for the many variables over which they have no control. It's not that actual traffic cannot fall outside the estimates, but between high and low, they cover 75% of likely outcomes. Otherwise the range would be so big as to render the forecast useless.

Forecasts do not work well for small areas or specific locations, and they do not work well for any low volume modes of traffic. However, highways engineers often behave as if they do.

And yet somehow they've managed to overestimate car traffic on a national level repeatedly by so much that it's been below the "low" estimate on that graphic. It looks an awful lot like the models are institutionally motorist, doesn't it?
MJR, mostly pedalling 3-speed roadsters. KL+West Norfolk BUG incl social easy rides http://www.klwnbug.co.uk
All the above is CC-By-SA and no other implied copyright license to Cycle magazine.

Vorpal
Moderator
Posts: 16305
Joined: 19 Jan 2009, 3:34pm
Location: Not there ;)

Re: Accuracy of historical bicycle miles travelled and cyclist casualty rates.

Postby Vorpal » 8 Jan 2019, 8:29pm

mjr wrote:
Vorpal wrote:Traffic forecast models are another thing altogether. The models that create forecasts are different to the models that estimate daily flow. completely different software is used.

Yes, but if the models that estimate daily flow are being fed into the models that create forecasts (because the actual traffic counts don't happen most years), then they both have effects.
That is true, and the estimates are used in the models. On a systemic basis, that should be okay. On a location by location basis, they should take manual counts, especially if it's for a new junction or something.

mjr wrote:And yet somehow they've managed to overestimate car traffic on a national level repeatedly by so much that it's been below the "low" estimate on that graphic. It looks an awful lot like the models are institutionally motorist, doesn't it?

IMO, the main problem is that the leaders & politicians promote a motor-centric economic model, and they want forecasts that support that thinking. Ergo, traffic forecasts are designed to fit into that. That's not to say it's *not* institutionally motorist, but I'm not sure that's the main problem.

I have worked on traffic models a few times as independent analyst, and when it makes a difference to the investment required, they consistently *underestimate* motor traffic, or fail to account for new housing and industrial estates, etc.

Like I say, traffic models are like statistics. You can make them say anything.

Of course, if they don't let the analysts do their jobs, no one will take them with more than a pinch of salt, anyway.
“In some ways, it is easier to be a dissident, for then one is without responsibility.”
― Nelson Mandela, Long Walk to Freedom