Am I missing something?

[akirk]

and bad grammar...

speed is not an absolute - it is variable, from 0 to 200+ in road cars
so: 'where speed was a factor' = 100% of all situations
and is of course a totally irrelevant comment until you qualify the speed...

Alasdair

[Pete317]

Esceptico said:

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.

I don't need lecturing on the meaning of words and phrases, thank you.

However, the subject of this thread is what, if anything, do the statistics tell us about speed and accident rates.
That's real world statistics, and your 'all else remaining equal' doesn't come into it, because, in the real world, things simply don't remain equal. If you try to isolate out individual factors, you very quickly find yourself with a whole bunch of tiny numbers which are too small to extract any significance from.

Nonetheless, if you wish to research something like the relationship between speed and accident rates, the first step is to develop models or hypotheses, quantify the effects according to these models/hypotheses, and finally to test your figures against either real-world statistics or, if that's impractical, something like computer simulations.

I'm happy to discuss this subject within such a framework, if you like.

Edited by Pete317 on Thursday 26th May 21:21

[Pete317]

tapereel said:

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

Wow! How have I missed seeing that for nearly 20 years!

If you wish to discuss it, how much time have you got?

[Esceptico]

Pete317 said:

tapereel said:

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

Wow! How have I missed seeing that for nearly 20 years!

If you wish to discuss it, how much time have you got?

Rather than post "happy to discuss it" why not actually discuss it by addressing the research paper referred to? If you don't think the research is valid then what were the faults with the methodology or statistical analysis?

[Esceptico]

Pete317 said:

However, the subject of this thread is what, if anything, do the statistics tell us about speed and accident rates.
That's real world statistics, and your 'all else remaining equal' doesn't come into it, because, in the real world, things simply don't remain equal. If you try to isolate out individual factors, you very quickly find yourself with a whole bunch of tiny numbers which are too small to extract any significance from.

Edited by Pete317 on Thursday 26th May 21:21

You keep posting this but never provide a scrap of evidence to support it. Why?

[Pete317]

Esceptico said:

Pete317 said:

tapereel said:

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

Wow! How have I missed seeing that for nearly 20 years!

If you wish to discuss it, how much time have you got?

Rather than post "happy to discuss it" why not actually discuss it by addressing the research paper referred to? If you don't think the research is valid then what were the faults with the methodology or statistical analysis?

Ok, in keeping with your 'all else remaining equal' theme, let's discuss the following bit:

Section 5.1.1 of the paper said:

However, even this comparison understates the importance of a 10 km/h difference in travelling speed. For example, consider two cars that are travelling side by side at a given instant, one travelling at 60 km/h and the other overtaking at 70 km/h. Suppose that a child runs onto the road at a point just beyond that at which the car travelling at 60 km/h can stop.
The other car will still be travelling at 45 km/h at that point. A difference in travelling speed of 10 km/h can mean a difference in impact speed of 45 km/h, or no impact and one at 45 km/h.

That, you might say, shows that, all else remaining equal, a 10 km/h difference in travelling speed results in a 45 km/h difference in impact speed - the implication being clearly that if you're driving 10 km/h faster in those conditions, then you'll hit whatever you hit at 45 km/h faster.

But that only holds true if both cars happen to be exactly alongside each other at that exact instant - bearing in mind that two cars travelling at different speeds can only be exactly alongside each other for an instant.
If, for example, they happen to be alongside each other a few metres further back, or a few metres further forward, or that exact instant at which the two cars are alongside each other changes by a small amount, the calculation of the impact speed changes profoundly.
So yes, all else being equal, the quoted relationship holds. But only if all else is equal at that precise quoted instant, when they cannot be equal at any other possible instant.
So which instant do we choose for all else to be equal?
And what makes that instant valid, and all the other possible instants invalid?
So what does that really tell us about the relationship between speed and accident risk, when just our choice of the instant at which all else is equal makes such a difference to the outcome?

If you really want to figure out the true relationship, you have to figure out a way to eliminate the effects of such 'cherry-picking'

Fortunately, that can be done - more tomorrow.

Edited by Pete317 on Thursday 26th May 22:53

[tapereel]

Pete317 said:

Esceptico said:

Pete317 said:

tapereel said:

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

Wow! How have I missed seeing that for nearly 20 years!

If you wish to discuss it, how much time have you got?

Rather than post "happy to discuss it" why not actually discuss it by addressing the research paper referred to? If you don't think the research is valid then what were the faults with the methodology or statistical analysis?

Ok, in keeping with your 'all else remaining equal' theme, let's discuss the following bit:

Section 5.1.1 of the paper said:

However, even this comparison understates the importance of a 10 km/h difference in travelling speed. For example, consider two cars that are travelling side by side at a given instant, one travelling at 60 km/h and the other overtaking at 70 km/h. Suppose that a child runs onto the road at a point just beyond that at which the car travelling at 60 km/h can stop.
The other car will still be travelling at 45 km/h at that point. A difference in travelling speed of 10 km/h can mean a difference in impact speed of 45 km/h, or no impact and one at 45 km/h.

That, you might say, shows that, all else remaining equal, a 10 km/h difference in travelling speed results in a 45 km/h difference in impact speed - the implication being clearly that if you're driving 10 km/h faster in those conditions, then you'll hit whatever you hit at 45 km/h faster.

But that only holds true if both cars happen to be exactly alongside each other at that exact instant - bearing in mind that two cars travelling at different speeds can only be exactly alongside each other for an instant.
If, for example, they happen to be alongside each other a few metres further back, or a few metres further forward, or that exact instant at which the two cars are alongside each other changes by a small amount, the calculation of the impact speed changes profoundly.
So yes, all else being equal, the quoted relationship holds. But only if all else is equal at that precise quoted instant, when they cannot be equal at any other possible instant.
So which instant do we choose for all else to be equal?
And what makes that instant valid, and all the other possible instants invalid?
So what does that really tell us about the relationship between speed and accident risk, when just our choice of the instant at which all else is equal makes such a difference to the outcome?

If you really want to figure out the true relationship, you have to figure out a way to eliminate the effects of such 'cherry-picking'

Fortunately, that can be done - more tomorrow.

Edited by Pete317 on Thursday 26th May 22:53

That argument is utter nonsense. The cars need not even be considered on the same day, week or year, simply compared in the same circumstance.
After reading your text for some months it is clear that you either deliberately obfuscate or are simply confused. Either way you are adding nothing to the debate on this subject and have nothing of note to oppose the two referenced reports. Unless you have evidence to oppose the findings in the report you have nothing to add. You have not evidenced the knowledge to criticise the authors of those reports and hence your weak and confused rankings based on limited, out of context extracts do not undermine their conclusions.
A few posts further up this thread you were saying there is no evidenced based conclusions of this type, now you choose to undermine what you said didn't exist.
You can keep denying the facts but everyone sees through that.

[Pete317]

tapereel said:

Pete317 said:

Esceptico said:

Pete317 said:

tapereel said:

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

Wow! How have I missed seeing that for nearly 20 years!

If you wish to discuss it, how much time have you got?

Rather than post "happy to discuss it" why not actually discuss it by addressing the research paper referred to? If you don't think the research is valid then what were the faults with the methodology or statistical analysis?

Ok, in keeping with your 'all else remaining equal' theme, let's discuss the following bit:

Section 5.1.1 of the paper said:

However, even this comparison understates the importance of a 10 km/h difference in travelling speed. For example, consider two cars that are travelling side by side at a given instant, one travelling at 60 km/h and the other overtaking at 70 km/h. Suppose that a child runs onto the road at a point just beyond that at which the car travelling at 60 km/h can stop.
The other car will still be travelling at 45 km/h at that point. A difference in travelling speed of 10 km/h can mean a difference in impact speed of 45 km/h, or no impact and one at 45 km/h.

That, you might say, shows that, all else remaining equal, a 10 km/h difference in travelling speed results in a 45 km/h difference in impact speed - the implication being clearly that if you're driving 10 km/h faster in those conditions, then you'll hit whatever you hit at 45 km/h faster.

But that only holds true if both cars happen to be exactly alongside each other at that exact instant - bearing in mind that two cars travelling at different speeds can only be exactly alongside each other for an instant.
If, for example, they happen to be alongside each other a few metres further back, or a few metres further forward, or that exact instant at which the two cars are alongside each other changes by a small amount, the calculation of the impact speed changes profoundly.
So yes, all else being equal, the quoted relationship holds. But only if all else is equal at that precise quoted instant, when they cannot be equal at any other possible instant.
So which instant do we choose for all else to be equal?
And what makes that instant valid, and all the other possible instants invalid?
So what does that really tell us about the relationship between speed and accident risk, when just our choice of the instant at which all else is equal makes such a difference to the outcome?

If you really want to figure out the true relationship, you have to figure out a way to eliminate the effects of such 'cherry-picking'

Fortunately, that can be done - more tomorrow.

Edited by Pete317 on Thursday 26th May 22:53

That argument is utter nonsense. The cars need not even be considered on the same day, week or year, simply compared in the same circumstance.
After reading your text for some months it is clear that you either deliberately obfuscate or are simply confused. Either way you are adding nothing to the debate on this subject and have nothing of note to oppose the two referenced reports. Unless you have evidence to oppose the findings in the report you have nothing to add. You have not evidenced the knowledge to criticise the authors of those reports and hence your weak and confused rankings based on limited, out of context extracts do not undermine their conclusions.
A few posts further up this thread you were saying there is no evidenced based conclusions of this type, now you choose to undermine what you said didn't exist.
You can keep denying the facts but everyone sees through that.

Really? Where exactly did I say that?
You're reading into my words exactly what you want to.

But back onto the subject - the point is that what I described is their hand-waving attempt at describing a mechanism to explain the stratospherically exponential nature of their findings.
And they cannot. There simply exists no plausible mechanism which would lead to a doubling of accidents, serious or otherwise, for a 5km/h increase in speed.
Taylor et al 2000 (TRL421) with their 1mph = 5% increase doesn't come remotely close to explaining it.
Neither does Elvik 2004 with his power model, who, incidentally, rejected Kloeden et al 1997 as unsuitable for inclusion in his study (reason: "Potentially systematic errors in measurement")

If you think that you can reconcile that, good luck.

Oh, and do leave the ad homs at the door next time.

ETA: links to the above: http://www.trg.dk/elvik/740-2004.pdf and http://www.trl.co.uk/reports-publications/download...

Edited by Pete317 on Friday 27th May 08:11

[tapereel]

So you just disagree with the evidence without credible evidence to do so. I think you are in denial.

[Esceptico]

Pete317 said:

Really? Where exactly did I say that?
You're reading into my words exactly what you want to.

But back onto the subject - the point is that what I described is their hand-waving attempt at describing a mechanism to explain the stratospherically exponential nature of their findings.
And they cannot. There simply exists no plausible mechanism which would lead to a doubling of accidents, serious or otherwise, for a 5km/h increase in speed.
Taylor et al 2000 (TRL421) with their 1mph = 5% increase doesn't come remotely close to explaining it.
Neither does Elvik 2004 with his power model, who, incidentally, rejected Kloeden et al 1997 as unsuitable for inclusion in his study (reason: "Potentially systematic errors in measurement")

If you think that you can reconcile that, good luck.

Oh, and do leave the ad homs at the door next time.

ETA: links to the above: http://www.trg.dk/elvik/740-2004.pdf and http://www.trl.co.uk/reports-publications/download...

Edited by Pete317 on Friday 27th May 08:11

Thanks for the link to the Norwegian report but why link to something that provides conclusive evidence to demolish your our arguments? I admit I only read the summary of the report to see the conclusions rather than all 148 pages but it says that a meta analysis of around 170 odd other studies showed clear evidence for the correlation of speed with accident rates, that the strength of relationship and consistency over so many studies indicated that the relationship was almost definitely causal and that the power model was supported (with some modfications). Some limitations of the studies were noted but not sufficient to overturn confidence in the conclusions of the study. What exactly is your argument then that speed and accident rate are not related?

[RobinOakapple]

Pete's argument continues to boil down to "driving faster means you've already passed the potential hazard by the time it occurs".

And is therefore nonsense.

[tapereel]

I submitted this many years ago to give some insight into the issue.

Think of 20 people in a boxing ring.
Tell them that they must walk around it at a slow walking pace and they must not touch any other person.
At the slow pace this is achieved.
Now tell them to walk as fast as they can without running. Observe the number of collisions.
Now they must run. Observe the collision rate and now how many sustain bruising.
Now get half to run and the other half to run but half the time stop or walk slowly. Count the collision rate and record the level of injuries.

Do you really need to conduct the experiment to see what happens?

If you want to simulate traffic more closely, insert a number of marked pathways on the floor and get the people to follow set routes at varying speeds.

Good luck with that 317, post up your results table.

[tapereel]

RobinOakapple said:

Pete's argument continues to boil down to "driving faster means you've already passed the potential hazard by the time it occurs".

And is therefore nonsense.

If you use the "driving faster...etc" argument then a career in comedy is for you. Logic is simply not your game. I'm afraid 317 is heading for the stage.

[tapereel]

Pete317 said:

Really? Where exactly did I say that?
You're reading into my words exactly what you want to.

But back onto the subject - the point is that what I described is their hand-waving attempt at describing a mechanism to explain the stratospherically exponential nature of their findings.
And they cannot. There simply exists no plausible mechanism which would lead to a doubling of accidents, serious or otherwise, for a 5km/h increase in speed.
Taylor et al 2000 (TRL421) with their 1mph = 5% increase doesn't come remotely close to explaining it.
Neither does Elvik 2004 with his power model, who, incidentally, rejected Kloeden et al 1997 as unsuitable for inclusion in his study (reason: "Potentially systematic errors in measurement")

If you think that you can reconcile that, good luck.

Oh, and do leave the ad homs at the door next time.

ETA: links to the above: http://www.trg.dk/elvik/740-2004.pdf and http://www.trl.co.uk/reports-publications/download...

Edited by Pete317 on Friday 27th May 08:11

No ad hom I'm afraid, I'm arguing against your position not you and I have said why. You can't even get that right. Again, just for you, no ad hom as I have argued against your position...not you.


Edited to add: I think you are missing something. That may or may not be ad hom.

[bryan35]

Maybe drifted off a little. What I was pointing out is the apparent gulf between the amount of speeding and the number of accidents. The amount of speeding is literally astronomic. 95 billion miles is more than 1000 astronomic units - to the sun and back 500 times. It doesn't fit easily into the human mind it's so large. Whereas people in accident would fit in a sports hall, and the number killed in a village hall. Speeding in itself is therefore very very clearly not dangerous in itself otherwise we'd see an absolute massacre on the roads. We don't.

[Dr Jekyll]

bryan35 said:

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

A few years back the phrase used was 'speed related' which included all accidents involving someone exceeding the limit, all accidents caused by driving too fast but within the limit, and all accidents involving factors deemed to be speed related. These factors included overtaking, driver being in a hurry, aggression, slippery roads and misjudging the speed of others. This is what they used to support the claim of '1/3 of accidents caused by speed'. The USA notoriously got the same 1/3 figure by the same method, except that they also included speeds of below posted minimums as a speed related factor.

It would also be interesting to know what is meant by 'seriously injured'. This used to include shock and concussion. One police force told their officers to stop recording these conditions among accident victims without medical confirmation, then hailed the reduction in serious injuries as due to speed cameras.

[Esceptico]

bryan35 said:

Maybe drifted off a little. What I was pointing out is the apparent gulf between the amount of speeding and the number of accidents. The amount of speeding is literally astronomic. 95 billion miles is more than 1000 astronomic units - to the sun and back 500 times. It doesn't fit easily into the human mind it's so large. Whereas people in accident would fit in a sports hall, and the number killed in a village hall. Speeding in itself is therefore very very clearly not dangerous in itself otherwise we'd see an absolute massacre on the roads. We don't.

190'000 people injured last year. Rather more than a sports hall. If you included all the people involved in accidents where no physical injury I suspect it would be many times that figure (someone drove into my car last year and there were 5 of us across two cars but no injuries). Reducing accidents is not just about reducing physical injuries, although that is the most important consideration.

[bryan35]

Esceptico said:

bryan35 said:

Maybe drifted off a little. What I was pointing out is the apparent gulf between the amount of speeding and the number of accidents. The amount of speeding is literally astronomic. 95 billion miles is more than 1000 astronomic units - to the sun and back 500 times. It doesn't fit easily into the human mind it's so large. Whereas people in accident would fit in a sports hall, and the number killed in a village hall. Speeding in itself is therefore very very clearly not dangerous in itself otherwise we'd see an absolute massacre on the roads. We don't.

190'000 people injured last year. Rather more than a sports hall. If you included all the people involved in accidents where no physical injury I suspect it would be many times that figure (someone drove into my car last year and there were 5 of us across two cars but no injuries). Reducing accidents is not just about reducing physical injuries, although that is the most important consideration.

3064 according to government figures, which is more or less a sports hall.
I'm not intending to argue the point either way. All I'm highlighting is the amount of speeding versus the amount of speeding related incidents. But even if you take all accidents and said every one was related to speeding, that's still 1 accident every 500,000 miles of speeding.

[Mill Wheel]

tapereel said:

I submitted this many years ago to give some insight into the issue.

Think of 20 people in a boxing ring.
Tell them that they must walk around it at a slow walking pace and they must not touch any other person.
At the slow pace this is achieved.
Now tell them to walk as fast as they can without running. Observe the number of collisions.
Now they must run. Observe the collision rate and now how many sustain bruising.
Now get half to run and the other half to run but half the time stop or walk slowly. Count the collision rate and record the level of injuries.

Do you really need to conduct the experiment to see what happens?

If you want to simulate traffic more closely, insert a number of marked pathways on the floor and get the people to follow set routes at varying speeds.

Good luck with that 317, post up your results table.

In theory, you would expect collisions and injuries - but you have fallen into the trap of IMPOSING a set of conditions into your theory, that in practice would NOT occur, if each individual were left to decide upon their own course and speed... in which case it would probably work out more like this:
https://youtu.be/GErEbUXCpiM
How many collisions do you count there?

[singlecoil]

Mill Wheel said:

In theory, you would expect collisions and injuries - but you have fallen into the trap of IMPOSING a set of conditions into your theory, that in practice would NOT occur, if each individual were left to decide upon their own course and speed... in which case it would probably work out more like this:
https://youtu.be/GErEbUXCpiM
How many collisions do you count there?

How many people did you see driving quickly in that video?

Your humorous example has worked against you.