Cyclist fatalities surge

User avatar
biker jk
Posts: 7001
Joined: Tue Nov 17, 2009 6:18 pm
Location: Sydney

Cyclist fatalities surge

Postby biker jk » Mon Aug 06, 2018 4:10 pm

45 deaths in the year to June 2018 versus 25 a year ago. Admittedly, the year ago figure was significantly below average. The 10-year average is 35.8, so 2018 year-to-date shows a rise of 25%.

https://goo.gl/dbRwkn

https://bitre.gov.au/statistics/safety/

http://roadsafety.gov.au/performance/ro ... -user.aspx

User avatar
Thoglette
Posts: 6605
Joined: Thu Feb 19, 2009 1:01 pm

Re: Cyclist fatalities surge

Postby Thoglette » Mon Aug 06, 2018 5:02 pm

And what's the standard deviation? Are we still still within normal process control limits, to borrow a phrase?

Otherwise we're chicken littling.
Stop handing them the stick! - Dave Moulton
"People are worthy of respect, ideas are not." Peter Ellerton, UQ

Jmuzz
Posts: 631
Joined: Mon Jan 29, 2018 1:42 pm

Re: Cyclist fatalities surge

Postby Jmuzz » Mon Aug 06, 2018 5:22 pm

Laws and infrastructure and other factors such as fuel prices which encourages more cycling is going to come with higher statistics.

Statistics really need to be presented with per * hours, per * kms to give an accurate picture otherwise it makes cycling look really bad.

If there is 100 fold increase in cycling, which is realistic, everyone has to be prepared to see a rise in statistics.

User avatar
Thoglette
Posts: 6605
Joined: Thu Feb 19, 2009 1:01 pm

Re: Cyclist fatalities surge

Postby Thoglette » Mon Aug 06, 2018 5:39 pm

<mode="cynic">
Clearly we need to ramp up Operations Pedro and Apollo to ensure cyclists are behaving safely!
</mode>
Stop handing them the stick! - Dave Moulton
"People are worthy of respect, ideas are not." Peter Ellerton, UQ

Jmuzz
Posts: 631
Joined: Mon Jan 29, 2018 1:42 pm

Re: Cyclist fatalities surge

Postby Jmuzz » Mon Aug 06, 2018 6:32 pm

The states will join forces with operation Appalling Pedo and save all bicycles from themselves!

mangoman
Posts: 13
Joined: Fri Feb 16, 2018 6:45 pm

Re: Cyclist fatalities surge

Postby mangoman » Mon Aug 06, 2018 10:42 pm

Agree with what jmuzz says, hopefully the number of cyclists is increasing.

Scintilla
Posts: 325
Joined: Sun Feb 25, 2018 5:36 pm

Re: Cyclist fatalities surge

Postby Scintilla » Mon Aug 06, 2018 11:24 pm

Jmuzz wrote:Laws and infrastructure and other factors such as fuel prices which encourages more cycling is going to come with higher statistics.
Record low fuel prices, crap road laws, lousy traffic conditions generally. all point to bicycling detterence.
Jmuzz wrote:Statistics really need to be presented with per * hours, per * kms to give an accurate picture otherwise it makes cycling look really bad.

If there is 100 fold increase in cycling, which is realistic, everyone has to be prepared to see a rise in statistics.
Our inadequate statistical data does not record such things as bicycle distance travelled. It's not important, you know!

Taking the National Cycling Strategy's recent data that has showed a steady decline in bicycle ridership from 2011 to 2017, we could presume that the general data of less than 1% bicycle mode-share is probably still the case and not increasing any significant amount.

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Tue Aug 07, 2018 7:41 am

As usual with anything statistically related, the reporting in news is poor.

This is the 30 year trend in annual cycling road fatalities:

Image

This is the same period but covering just the January to June period for each year:

Image

The average for the first 6 months of the year are:
Last 10 years: 21.80 fatalities
Last 20 years: 22.15 fatalities

So very little change in the absolute numbers.

On a per km travelled basis one might be able to make some additional assessment.

Cyclophiliac
Posts: 1014
Joined: Thu Apr 10, 2014 7:48 am

Re: Cyclist fatalities surge

Postby Cyclophiliac » Tue Aug 07, 2018 8:16 am

Alex Simmons/RST wrote:As usual with anything statistically related, the reporting in news is poor.

This is the 30 year trend in annual cycling road fatalities:

Image

This is the same period but covering just the January to June period for each year:

Image

The average for the first 6 months of the year are:
Last 10 years: 21.80 fatalities
Last 20 years: 22.15 fatalities

So very little change in the absolute numbers.

On a per km travelled basis one might be able to make some additional assessment.
Can you provide a link to the source data for the above? I'd like to be able to use it myself.

User avatar
g-boaf
Posts: 21317
Joined: Mon Sep 26, 2011 6:11 pm

Re: Cyclist fatalities surge

Postby g-boaf » Tue Aug 07, 2018 8:44 am

biker jk wrote:45 deaths in the year to June 2018 versus 25 a year ago. Admittedly, the year ago figure was significantly below average. The 10-year average is 35.8, so 2018 year-to-date shows a rise of 25%.

https://goo.gl/dbRwkn

https://bitre.gov.au/statistics/safety/

http://roadsafety.gov.au/performance/ro ... -user.aspx
Time to bring in harsher penalties for not riding with a helmet, ban riding on main roads and have weekly Operation Pedro stings.

Basically do everything to discourage cycling. If nobody rides bicycles, the statistics will be much better.

I'm in a cynical mood today.
Last edited by g-boaf on Tue Aug 07, 2018 10:49 am, edited 1 time in total.

AdelaidePeter
Posts: 1230
Joined: Wed Jun 07, 2017 11:13 am

Re: Cyclist fatalities surge

Postby AdelaidePeter » Tue Aug 07, 2018 10:29 am

Alex Simmons/RST wrote:As usual with anything statistically related, the reporting in news is poor.

This is the 30 year trend in annual cycling road fatalities:

Image

This is the same period but covering just the January to June period for each year:

Image

The average for the first 6 months of the year are:
Last 10 years: 21.80 fatalities
Last 20 years: 22.15 fatalities

So very little change in the absolute numbers.

On a per km travelled basis one might be able to make some additional assessment.
Your links don't work, but based on every time I've looked at this, I'm sure you're right. I'm all for reducing the bicycle toll, of course, but these numbers are almost certainly within normal statistical variation.

A quick tutorial on this sort of thing: for random independent events (like bicycle deaths), the standard deviation is the square root of the mean; and anything within 2 standard deviations of the mean is to be expected (very roughly speaking). Average (mean) number of bicycle deaths per year over the last 10 years (2008 to 2017) is 35.7, so let's round that to 36, that means the standard deviation is about 6, so we'd expect the annual total to be anywhere from 24 to 48. So it can vary by a factor of 2 year to year simply due to random effects, because the numbers are fairly small (though still too large, of course).

User avatar
Cycleops70
Posts: 402
Joined: Fri Jan 10, 2014 10:56 am
Location: Perth
Contact:

Re: Cyclist fatalities surge

Postby Cycleops70 » Tue Aug 07, 2018 10:34 am

g-boaf wrote:Time to bring in harsher penalties for not riding with a helmet, ban riding on main roads and have weekly Operation Pedro stings.

Basically do everything to discourage cycling. If nobody rides bicycles, the statistics will be much better.

I'm in a cynical today.
I recall some time ago Duncan Gay actually claim this in an interview.
When the interviewer challenged that his actions would discourage people from getting on a bike, he took credit for saving their life!
https://www.youtube.com/watch?v=EbvYkATHYMQ at 5:11

AdelaidePeter
Posts: 1230
Joined: Wed Jun 07, 2017 11:13 am

Re: Cyclist fatalities surge

Postby AdelaidePeter » Tue Aug 07, 2018 10:39 am

Cyclophiliac wrote:Can you provide a link to the source data for the above? I'd like to be able to use it myself.
I can make a plot of just about anything from the BITRE data, from 1989 onwards. What would you like? Ask and you shall receive (* probably, and not until I'm home).

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Tue Aug 07, 2018 1:47 pm

Cyclophiliac wrote:Can you provide a link to the source data for the above? I'd like to be able to use it myself.
https://bitre.gov.au/statistics/safety/ ... abase.aspx

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Tue Aug 07, 2018 2:20 pm

AdelaidePeter wrote:Your links don't work, but based on every time I've looked at this, I'm sure you're right. I'm all for reducing the bicycle toll, of course, but these numbers are almost certainly within normal statistical variation.
The image links? They appear OK for me but they are those extra long google photo links which seem to cause problems for some people to view them.

Not sure how else to put images here since the great Photobucket debacle.

Here are direct links:
https://photos.app.goo.gl/rmjQqetKLc8KPfVL7
https://photos.app.goo.gl/8orWZofLqtsJ9g2B8

I usually list my data source on charts, this was a quick and dirty.
AdelaidePeter wrote:A quick tutorial on this sort of thing: for random independent events (like bicycle deaths), the standard deviation is the square root of the mean; and anything within 2 standard deviations of the mean is to be expected (very roughly speaking). Average (mean) number of bicycle deaths per year over the last 10 years (2008 to 2017) is 35.7, so let's round that to 36, that means the standard deviation is about 6, so we'd expect the annual total to be anywhere from 24 to 48. So it can vary by a factor of 2 year to year simply due to random effects, because the numbers are fairly small (though still too large, of course).
For the 10 years 2008-2017, the average annual number of cyclist fatalities was 35.8 with a standard deviation of 7.2.

For the first six months of each year over the same period, the average is 21.0 with an SD of 4.9.
Which means the average for Jul-Dec is 14.8 fatalities or 70% of the Jan-Jun period.
Why? because the numbers drop a lot in the coldest months of Jul-Aug-Sep.

In 2018 June YTD there have been 21 fatalities, smack bang on the average for the previous 10 years.


For the original linked News article to suggest there has been an 80% increase in cycling fatalities is nothing short of really poor reporting, no matter how the numbers are cut and diced. It's just wrong.

Why is it wrong? Because it does not:
i. compare data from the same time period, it's cherry picking; nor
ii. place the data in context with longer term trends.

In the case of i., the long term trends show that, on average, 59% of cyclist fatalities occur in the first half of the year.

What worse in this news report is that 2017 was slightly unusual in that cyclist fatalities were higher in the second half of the year - 54% compared with the 10 year average of 41%.

Yes we all want the situation for vulnerable road users to improve, but poor use of numbers and made up statistics does not help the cause.

Let use data that portrays reality and helps understand the true underlying causes.
Last edited by Alex Simmons/RST on Tue Aug 07, 2018 2:32 pm, edited 1 time in total.

AdelaidePeter
Posts: 1230
Joined: Wed Jun 07, 2017 11:13 am

Re: Cyclist fatalities surge

Postby AdelaidePeter » Tue Aug 07, 2018 2:32 pm

Alex Simmons/RST wrote:
AdelaidePeter wrote:
AdelaidePeter wrote:A quick tutorial on this sort of thing: for random independent events (like bicycle deaths), the standard deviation is the square root of the mean; and anything within 2 standard deviations of the mean is to be expected (very roughly speaking). Average (mean) number of bicycle deaths per year over the last 10 years (2008 to 2017) is 35.7, so let's round that to 36, that means the standard deviation is about 6, so we'd expect the annual total to be anywhere from 24 to 48. So it can vary by a factor of 2 year to year simply due to random effects, because the numbers are fairly small (though still too large, of course).
For the 10 years 2008-2017, the average annual number of cyclist fatalities was 35.8 with a standard deviation of 7.2.

For the first six months of each year over the same period, the average is 21.0 with an SD of 4.9.
Just clarifying: the square root of the mean is the theoretical number for what you would expect the SD to be, if the events are truly independent and always at the same rate. The numbers you quote seem to back this up as a reasonable assumption.

The rest of what you say is a good analysis. If they really are comparing year-to-date with complete years, that is incredibly sloppy.

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Tue Aug 07, 2018 2:37 pm

AdelaidePeter wrote:
Alex Simmons/RST wrote:
AdelaidePeter wrote:
For the 10 years 2008-2017, the average annual number of cyclist fatalities was 35.8 with a standard deviation of 7.2.

For the first six months of each year over the same period, the average is 21.0 with an SD of 4.9.
Just clarifying: the square root of the mean is the theoretical number for what you would expect the SD to be, if the events are truly independent and always at the same rate. The numbers you quote seem to back this up as a reasonable assumption.

The rest of what you say is a good analysis. If they really are comparing year-to-date with complete years, that is incredibly sloppy.
Yep, I was just giving the actual numbers to compare with the back of the envelope estimate.

march83
Posts: 1046
Joined: Mon Apr 29, 2013 10:41 pm

Re: Cyclist fatalities surge

Postby march83 » Tue Aug 07, 2018 2:41 pm

@alex, I suggest using Imgur. Super easy to use, free, no membership required, etc

User avatar
MichaelB
Posts: 14775
Joined: Tue Feb 20, 2007 9:29 am
Location: Adelaide, South Australia

Re: Cyclist fatalities surge

Postby MichaelB » Tue Aug 07, 2018 2:46 pm

g-boaf wrote:
Time to bring in harsher penalties for not riding with a helmet, ban riding on main roads and have weekly Operation Pedro stings.

Basically do everything to discourage cycling. If nobody rides bicycles, the statistics will be much better.

I'm in a cynical mood today.
You forgot cycling registration and compulsory high vis jackets ....

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Wed Aug 08, 2018 7:00 am

march83 wrote:@alex, I suggest using Imgur. Super easy to use, free, no membership required, etc
OK. Testing 123...

Image

Hmmm, that didn't seem to work.

Image

That didn't work either.

Not sure about this imgur option.

User avatar
andrewjcw
Posts: 197
Joined: Wed Jan 21, 2015 12:56 pm
Location: Newcastle, NSW

Re: Cyclist fatalities surge

Postby andrewjcw » Wed Aug 08, 2018 7:10 am

You need to link the actual img url, not the page the image is hosted on. The proper image url will end with .png or .jpg. You can find it on most images by right click -> copy link address or right click -> open in new tab (and then copy that url). So upload it as you have done to imgur, and then copy the image url out of the 'finished' page imgur gives you.

Image
https://www.strava.com/athletes/andrewjcw" onclick="window.open(this.href);return false;

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Wed Aug 08, 2018 12:59 pm

andrewjcw wrote:You need to link the actual img url, not the page the image is hosted on. The proper image url will end with .png or .jpg. You can find it on most images by right click -> copy link address or right click -> open in new tab (and then copy that url). So upload it as you have done to imgur, and then copy the image url out of the 'finished' page imgur gives you.
I can't even find the image on my imgur page. It exists as I put one there but no idea where it is.

¯\_(ツ)_/¯

AdelaidePeter
Posts: 1230
Joined: Wed Jun 07, 2017 11:13 am

Re: Cyclist fatalities surge

Postby AdelaidePeter » Wed Aug 08, 2018 1:43 pm

Alex Simmons/RST wrote: For the original linked News article to suggest there has been an 80% increase in cycling fatalities is nothing short of really poor reporting, no matter how the numbers are cut and diced. It's just wrong.
Actually I've noticed that it is correct by one "dicing" of the data. If you take financial years (July through to June), then there were indeed 25 fatalities in 2016-17 and 45 fatalities in 2017-18. But I would argue that it's some rather selective dicing.

What the article omits (as we've both been pointing out) is that the numbers are all over the place and 45 is well within what we'd expect with statistical variation. From 2012-13 to 2017-18, the numbers for the last 6 financial years have been: 38, 58, 30, 35, 25, 45.

[Actually that 58 does look unusual: it is more than 3 SDs above the average of 36, so maybe there was something about that summer, since the warmer months are worst for cycling fatalities. In fact it's the highest number since 1990-1991, and the only one above 50 since 1996-97. But the 45, while sickeningly high, is well within what we'd expect].

Of course, the numbers are way too high and I applaud any effort to get them down. It's just the misuse of statistics which annoys me.

User avatar
Thoglette
Posts: 6605
Joined: Thu Feb 19, 2009 1:01 pm

Re: Cyclist fatalities surge

Postby Thoglette » Wed Aug 08, 2018 1:56 pm

Also interesting article in The Guardian today. Will link later, unless someone beats me to it.
Stop handing them the stick! - Dave Moulton
"People are worthy of respect, ideas are not." Peter Ellerton, UQ

User avatar
Alex Simmons/RST
Expert
Posts: 4997
Joined: Fri May 02, 2008 3:51 pm
Contact:

Re: Cyclist fatalities surge

Postby Alex Simmons/RST » Wed Aug 08, 2018 4:49 pm

AdelaidePeter wrote:
Alex Simmons/RST wrote: For the original linked News article to suggest there has been an 80% increase in cycling fatalities is nothing short of really poor reporting, no matter how the numbers are cut and diced. It's just wrong.
Actually I've noticed that it is correct by one "dicing" of the data. If you take financial years (July through to June), then there were indeed 25 fatalities in 2016-17 and 45 fatalities in 2017-18. But I would argue that it's some rather selective dicing.

What the article omits (as we've both been pointing out) is that the numbers are all over the place and 45 is well within what we'd expect with statistical variation. From 2012-13 to 2017-18, the numbers for the last 6 financial years have been: 38, 58, 30, 35, 25, 45.

[Actually that 58 does look unusual: it is more than 3 SDs above the average of 36, so maybe there was something about that summer, since the warmer months are worst for cycling fatalities. In fact it's the highest number since 1990-1991, and the only one above 50 since 1996-97. But the 45, while sickeningly high, is well within what we'd expect].

Of course, the numbers are way too high and I applaud any effort to get them down. It's just the misuse of statistics which annoys me.
This is the trend by financial years:

Image

The most recent 10 financial years average is 36.6 with SD of 9.3 fatalities.
The most recent 20 financial years average is 37.1 with SD of 7.7 fatalities.

IOW the number of fatalities in 2017-18 financial year is within one standard deviation of the 10 year average and all but within 1SD of the 20 year average.

Who is online

Users browsing this forum: find_bruce