Am I missing something?

Author
Discussion

bryan35

Original Poster:

1,906 posts

240 months

Wednesday 25th May 2016
quotequote all
In any scientific study of cause and effect you generally need a large sample size ideally over a large time scale. Positive results are always large, clear and repeatable. e.g. there's clear indications that smoking can be demonstrated to cause serious health problems, whereas there's little evidence that mobile phones cause brain tumours. That said some people get telegrams from the Queen having smoked all their lives, and people do get cranial problems where phones are held, but these are hardly the norm.

So I was looking at stats that are available on the net about the dreaded subject of speeding.

http://think.direct.gov.uk/speed.html

This government site says that in 2013 '3,064 people were killed or seriously injured in crashes where speed was a factor'

Fair enough, it's not clear whether these were actual speeding instances, but if we were to just group 'speeders' with people that generally travel too fast for the conditions it'll maybe save time.

So how prevalent is breaking the speed limit?

with reference to this government site...

https://www.gov.uk/government/uploads/system/uploa...


...we see that rather a lot of vehicles break the speed limit. We see numbers as high as 46% and 47% for motorways and built up areas. Lastly, how many miles were actually driven in 2013?

https://www.gov.uk/government/uploads/system/uploa...

the answer is 303700000000 - 303.7 billion.

that's quite a lot of miles driven over the speed limit, maybe 90-100 million when you run the numbers.

dividing this by the KSI number mentioned earlier (3064), statistically you could expect a KSI every 31 million miles of speeding. To put that into perspective, if you drive 10000 miles a year and were to drive all those miles over the speed limit, you could expect a KSI roughly every 3,100 years.


So there it is. Using mobile phones causes brain tumours. - I think.






RobinOakapple

2,802 posts

111 months

Wednesday 25th May 2016
quotequote all
Yes

What you are missing is a thread title.

jdw100

4,069 posts

163 months

Wednesday 25th May 2016
quotequote all
You would also need to factor in what speed over the limit people are doing.

If I drive my 10,000 miles a year at, say, 5mph over the speed limit I'd say risk is quite low. If I drive at 50mph over the speed limit all the time I'd expect to be dying or killing people more often than every 3,000 years, probably once a day at least I would think!

Pete317

1,430 posts

221 months

Wednesday 25th May 2016
quotequote all
jdw100 said:
If I drive at 50mph over the speed limit all the time I'd expect to be dying or killing people more often than every 3,000 years, probably once a day at least I would think!
Meanwhile, back on Planet Earth ...

StuTheGrouch

5,714 posts

161 months

Wednesday 25th May 2016
quotequote all
bryan35 said:
In any scientific study of cause and effect you generally need a large sample size ideally over a large time scale. Positive results are always large, clear and repeatable. e.g. there's clear indications that smoking can be demonstrated to cause serious health problems, whereas there's little evidence that mobile phones cause brain tumours. That said some people get telegrams from the Queen having smoked all their lives, and people do get cranial problems where phones are held, but these are hardly the norm.

So I was looking at stats that are available on the net about the dreaded subject of speeding.

http://think.direct.gov.uk/speed.html

This government site says that in 2013 '3,064 people were killed or seriously injured in crashes where speed was a factor'

Fair enough, it's not clear whether these were actual speeding instances, but if we were to just group 'speeders' with people that generally travel too fast for the conditions it'll maybe save time.

So how prevalent is breaking the speed limit?

with reference to this government site...

https://www.gov.uk/government/uploads/system/uploa...


...we see that rather a lot of vehicles break the speed limit. We see numbers as high as 46% and 47% for motorways and built up areas. Lastly, how many miles were actually driven in 2013?

https://www.gov.uk/government/uploads/system/uploa...

the answer is 303700000000 - 303.7 billion.

that's quite a lot of miles driven over the speed limit, maybe 90-100 million when you run the numbers.

dividing this by the KSI number mentioned earlier (3064), statistically you could expect a KSI every 31 million miles of speeding. To put that into perspective, if you drive 10000 miles a year and were to drive all those miles over the speed limit, you could expect a KSI roughly every 3,100 years.


So there it is. Using mobile phones causes brain tumours. - I think.


What is that 3,064 as a percentage of all driving-related deaths and serious injuries?

As a previous poster allured to, one would need to analyse each of those 3,064 to determine how far over the speed limit each case was. You also need to allow for other joint factors (i.e. over the speed limit by X% margin AND over the legal alcohol limit). A simple spreadsheet and stats analysis would allow this.

All your analysis has done is test a very basic hypothesis of 'speeding causes deaths and serious injury' (this is proven). The next step, if you're bored enough to do this, is to test the hypothesis that 'speeding in excess of X% over the legal limit, in the absence of other factors, is the cause of a statistically significant proportion of all driving-related deaths and injuries'. You need a lot of data for that, over a number of years. I suspect publicly available data would not suffice.



Osinjak

5,453 posts

120 months

Wednesday 25th May 2016
quotequote all
StuTheGrouch said:
All your analysis has done is test a very basic hypothesis of 'speeding causes deaths and serious injury' (this is proven). The next step, if you're bored enough to do this, is to test the hypothesis that 'speeding in excess of X% over the legal limit, in the absence of other factors, is the cause of a statistically significant proportion of all driving-related deaths and injuries'. You need a lot of data for that, over a number of years. I suspect publicly available data would not suffice.
Hmm. Adenoids on. Surely you would test the null hypothesis 'speeding in excess of xxxxx is not the cause of xxxx' through the calculation of a p-value? In and of itself the p-value does not indicate causation, one would ordinarily conduct regression analysis. Even then it's not definitive. Adenoids off.

bryan35

Original Poster:

1,906 posts

240 months

Wednesday 25th May 2016
quotequote all
well, I remember seeing a publicity photo for Humberside safety camera partnership that had a picture of a speed camera and the slogan 'Catching Killers', plus of course a line from a chief constable saying 'drifting over the speed limit is no different to drifting a knife into someone'. Both specifically say that going over the speed limit IN ITSELF kills. However, considering the colossal number of miles driven over the speed limit (approx 95,000,000,000) it can't just be good luck that the accidents are so incredibly low.

Pete317

1,430 posts

221 months

Wednesday 25th May 2016
quotequote all
They love bandying around essentially meaningless statistics.
Truth is, there are so many factors, a lot of them random, which have a far more profound effect on accidents, as well as casualties, than a few mph of travelling speed does, that it would take a lot of data gathered over a very long time on a global scale, before any effect whatsoever can be detected.

Jasandjules

69,825 posts

228 months

Wednesday 25th May 2016
quotequote all
It is also "a factor". That means there are other factors. Which is the primary factor however is clearly a question as only in those cases where speeding is the primary factor does it count. This also begs the question as to whether excess speed is "speeding" or whether or not excess speed can be for example 50mph in heavy rain on an NSL road.....

Esceptico

7,334 posts

108 months

Thursday 26th May 2016
quotequote all
A few points...

Both the anti and pro speeding brigades misuse and misinterpret statistics to try to support their own positions. I'm always wary of people with agendas as their arguments are all aimed at proving or supporting a position they have taken for ideological purposes. Saying that "speedig kills" is emotive nonsense because most of us speed every day. However, higher speeds (and speeding) will have an impact on accident rates and severity of crashes - all else being equal. Of course, all else is not equal because cars and roads are being made safer all the time and there are a host of other factors (reduction in drink driving, changing demographics, etc) that mean that proving any simple relationship between speeding and accidents is difficult enough that both sides can claim that speed has little or lots of impact on the number of accidents.

The pro speeding bridge love to focus on the low number of deaths. They seem to ignore the 180'000 odd people that are injured nor the hundreds of thousands of proper involved in RTAs that don't cause physical injury but are still distressing and cause economic pain. Reducing accident rates is not all about avoiding injury and death. I would welcome safer roads and better drivers just to cut my insurance premiums and my tax bills.


Pete317

1,430 posts

221 months

Thursday 26th May 2016
quotequote all
...and you've just taken it as a given that reducing speed will reduce accident rates

Esceptico

7,334 posts

108 months

Thursday 26th May 2016
quotequote all
Pete317 said:
...and you've just taken it as a given that reducing speed will reduce accident rates
I am open to be persuaded by evidence that speed limits and average speeds are not related to accident rates but would require strong evidence because it is an exceptional claim. Human error causes most accidents. Factors that increase the risk of people making an error or turning errors into actual crashes would be expected to increase the accident rate. Alcohol impairs people's judgement, attitude to risk and motor skills so it would be very surprising if drink driving didn't affect accident rates. Similarly, experience and attitude to risk would at face value be likely to increase errors so I would be surprised if young and new drivers didn't have a higher accident rate. Same for speed - faster you go the less time you have to deal with the unexpected and the more difficult it will be to take action to avoid an incident and the bigger the impact if you do crash. For risk not to increase with speed (all else being equal - I don't know if I should write that as many on here don't seem to understand what it actually means) there would have to be some other compensatory mechanism at work. That would have to be proved.

A bit like climate change. How radiation from the earth is absorbed by gases in the atmosphere and then re-radiated (thereby preventing the loss of some heat to space) is basic physics. Without any greenhouse gases the global temperature would be about 30 degrees less than it is. Therefore a base position to take is that increasing greenhouse gases in the atmosphere would over time increase global temperatures unless there was a compensating mechanism that lead to negative feedback (eg increased temperatures leads to higher cloud cover and a greater albedo effect). Such compensating effects couldn't be assumed though by anyone being objective.


StuTheGrouch

5,714 posts

161 months

Thursday 26th May 2016
quotequote all
Osinjak said:
StuTheGrouch said:
All your analysis has done is test a very basic hypothesis of 'speeding causes deaths and serious injury' (this is proven). The next step, if you're bored enough to do this, is to test the hypothesis that 'speeding in excess of X% over the legal limit, in the absence of other factors, is the cause of a statistically significant proportion of all driving-related deaths and injuries'. You need a lot of data for that, over a number of years. I suspect publicly available data would not suffice.
Hmm. Adenoids on. Surely you would test the null hypothesis 'speeding in excess of xxxxx is not the cause of xxxx' through the calculation of a p-value? In and of itself the p-value does not indicate causation, one would ordinarily conduct regression analysis. Even then it's not definitive. Adenoids off.
Correct, however I didn't want to bring null hypotheses into it!

RobinOakapple

2,802 posts

111 months

Thursday 26th May 2016
quotequote all
Esceptico said:
I am open to be persuaded by evidence that speed limits and average speeds are not related to accident rates but would require strong evidence because it is an exceptional claim...
Pretty much my position too.

Pete317

1,430 posts

221 months

Thursday 26th May 2016
quotequote all
Esceptico said:
Pete317 said:
...and you've just taken it as a given that reducing speed will reduce accident rates
I am open to be persuaded by evidence that speed limits and average speeds are not related to accident rates but would require strong evidence because it is an exceptional claim. Human error causes most accidents.
Can you show that reducing speed will noticeably reduce accident rates, in the face of all the other factors involved - many of which have a profound effect on accident and casualty rates, and many of which are random and can't be controlled for?

As you said yourself, all other things are never equal - you just don't seem to grasp the scope of the unequal-ness.

Esceptico said:
- I don't know if I should write that as many on here don't seem to understand what it actually means.
Are you sure you're not one of those who don't seem to understand what it actually means?

RobinOakapple

2,802 posts

111 months

Thursday 26th May 2016
quotequote all
Pete317 said:
Esceptico said:
Pete317 said:
...and you've just taken it as a given that reducing speed will reduce accident rates
I am open to be persuaded by evidence that speed limits and average speeds are not related to accident rates but would require strong evidence because it is an exceptional claim. Human error causes most accidents.
Can you show that reducing speed will noticeably reduce accident rates, in the face of all the other factors involved - many of which have a profound effect on accident and casualty rates, and many of which are random and can't be controlled for?

As you said yourself, all other things are never equal - you just don't seem to grasp the scope of the unequal-ness.

Esceptico said:
- I don't know if I should write that as many on here don't seem to understand what it actually means.
Are you sure you're not one of those who don't seem to understand what it actually means?
Can you show that there was anything in that post other than contradiction?

Pete317

1,430 posts

221 months

Thursday 26th May 2016
quotequote all
RobinOakapple said:
Can you show that there was anything in that post other than contradiction?
Can you show anything of substance whatsoever?

Osinjak

5,453 posts

120 months

Thursday 26th May 2016
quotequote all
StuTheGrouch said:
Osinjak said:
StuTheGrouch said:
All your analysis has done is test a very basic hypothesis of 'speeding causes deaths and serious injury' (this is proven). The next step, if you're bored enough to do this, is to test the hypothesis that 'speeding in excess of X% over the legal limit, in the absence of other factors, is the cause of a statistically significant proportion of all driving-related deaths and injuries'. You need a lot of data for that, over a number of years. I suspect publicly available data would not suffice.
Hmm. Adenoids on. Surely you would test the null hypothesis 'speeding in excess of xxxxx is not the cause of xxxx' through the calculation of a p-value? In and of itself the p-value does not indicate causation, one would ordinarily conduct regression analysis. Even then it's not definitive. Adenoids off.
Correct, however I didn't want to bring null hypotheses into it!
biggrin

Esceptico

7,334 posts

108 months

Thursday 26th May 2016
quotequote all
Pete317 said:
Esceptico said:
Pete317 said:
...and you've just taken it as a given that reducing speed will reduce accident rates
I am open to be persuaded by evidence that speed limits and average speeds are not related to accident rates but would require strong evidence because it is an exceptional claim. Human error causes most accidents.
Can you show that reducing speed will noticeably reduce accident rates, in the face of all the other factors involved - many of which have a profound effect on accident and casualty rates, and many of which are random and can't be controlled for?

As you said yourself, all other things are never equal - you just don't seem to grasp the scope of the unequal-ness.

Esceptico said:
- I don't know if I should write that as many on here don't seem to understand what it actually means.
Are you sure you're not one of those who don't seem to understand what it actually means?
For avoidance of doubt here is a definition:


https://en.m.wikipedia.org/wiki/Ceteris_paribus

With respect to speed and driving it means that speed is the only variable so you have to compare different speeds on the same road under the same conditions (same weather, time of day, etc). Arguing that motorways have higher average speeds yet are safer than urban roads is irrelevant because you are not comparing like with like. Arguing that other factors (like driver training) also impact accident rates and therefore speed doesn't matter is also plain wrong. If you drop a leaf on a windy day, where it lands will be influenced by the wind but that doesn't disprove that the leaf falls because of gravity.


tapereel

1,860 posts

115 months

Thursday 26th May 2016
quotequote all
I think you guys are missing something! This work has already been done for you eliminating the need for pontification, wild speculation and hopeful bias.

See this report from 1997: http://casr.adelaide.edu.au/speed/SPEED-V1.PDF

Reanalysed here in 2002: https://infrastructure.gov.au/roads/safety/publica...

Enjoy and perhaps give up on your wild guesses. smile