Safe Speed Forums

Come on sort out your data dredging and selective reporting.

Where are your subjects on 2008, 2009, 2010 fatality stats? They don't seem to appear after your gleeful reports of 2006 and 7.

Where is your letter to Professor Allsop to explain away his finding of benefits of speed enforcement treatment; is that too risky for your unproven theories?

Where was the peer review of the Safespeed therum? Is there no peer?

Awaits denial and why I'm obviously off-beam again.

I suspect (well, I always have) that you haven't a clue what you're talking about, since your reply offered zero substantiation, and rather appears to agree with me.

Using specific examples can be troublesome, but let me point out that this is a maths lesson for you, not an actual attempt at purporting figures for any particular site:

Suppose a stretch of road has had one collision which results in one person being seriously injured every 10 years for, lets say, 40 years. In order to calculate the mean (also known as the average, if that's a term you're more familiar with from your o-level maths) you divide the number of accidents over the period, 4, by the length of the period, 40, in order to produce a result of 0.1 KSIs per year.

Now for the next 9 years there are no accidents, but sure enough, in the tenth there is a collision which this time results in all 4 of the vehicle's occupants being seriously injured; here is where the fun with maths really gets going:

The mean for the final 3 years of the period is 1.33 (4/3)! (The mean for the entire period is now 0.16 (8/50))

A camera is installed and for the next 9 years there are no collisions, the SCP heralds the success of the camera and claims it has saved 12 lives (1.33*9)! The following year there is a collision and one person is seriously injured. There are mutterings about people knowing the location of the camera and "manipulating" it, that must be why it's effectiveness is waning.

For the next 40 years there is again only one serious injury every 10 years, so the means are as follows:

50 years prior to camera installation (disregarding anomalous year) 0.1
50 years prior to camera installation (including anomalous year) 0.16
3 years prior to camera installation 1.33
3 years post camera installation 0
50 years post camera installation 0.1
100 year period 0.13

So Greenshed, using your expert skills, what was the benefit of installing the camera in this example:

(a) Massive! It saved 66.5 lives in the 50 years after it's installation!!
(b) Huge! It reduced the mean number of KSIs from 1.33 per year to zero!
(c) Significant! It reduced the mean number of KSIs from 1.33 per year to 0.1 per year, a reduction of 93%!
(d) Debatable; it reduced the mean from 0.16 to 0.1, not necessarily statistically significant.
(e) Nil. It's quite clear that we'd expect the mean after the anomalous year to, at worst, stay the same as it was prior to it, regardless of any intervention.

_________________
Regulation without education merely creates more criminals.

Blimey Robin, it will be Xmas by the time he's digested half of that, he won't be around for another six months at least !

_________________
My views do not represent Safespeed but those of a driver who has driven for 39 yrs, in all conditions, at all times of the day & night on every type of road and covered well over a million miles, so knows a bit about what makes for safety on the road,what is really dangerous and needs to be observed when driving and quite frankly, the speedo is way down on my list of things to observe to negotiate Britain's roads safely, but I don't expect some fool who sits behind a desk all day to appreciate that.

Still if his his reading ability is far better than his maths, he should have at least learnt the meaning of another new word today..."mean" he just needs to work out what "average " means now...

_________________
My views do not represent Safespeed but those of a driver who has driven for 39 yrs, in all conditions, at all times of the day & night on every type of road and covered well over a million miles, so knows a bit about what makes for safety on the road,what is really dangerous and needs to be observed when driving and quite frankly, the speedo is way down on my list of things to observe to negotiate Britain's roads safely, but I don't expect some fool who sits behind a desk all day to appreciate that.

I think that should be sticky somewhere Robin. It's like a Dummies guide to RTTM

_________________
The views expressed in this post are personal opinions and do not necessarily represent the views of Safe Speed.
You will be branded a threat to society by going over a speed limit where it is safe to do so, and suffer the consequences of your actions in a way criminals do not, more so than someone who is a real threat to our society.

You mean like this little snippet from his paper?



Now I don't know where the good Prof received his education, but the implication that a vehicle would have ended up at the same place at the same time despite having travelled at a different speed, appears to fly in the face of all known laws of motion.

But of course you'll be able to explain that little paradox to us, wont you, GS?

_________________
Only when ideology, prejudice and dogma are set aside does the truth emerge - Kepler

A masterly exposition indeed. It should be engraved on velum. (No sarcasm intended)

_________________
When I see an adult on a bicycle, I do not despair for the future of the human race. H.G. Wells
When I see a youth in a motor car I do d.c.brown

Lancashire camera site accident stats now published in detail:

http://www.safe2travel.co.uk/more_info. ... rent_id=83

All statisticians should have a good time with these figures !

When you look at the detail in the figures, they must have been collecting these for years. Yet it is only when they are commanded to that they release the figures for public consumption. I think that speaks volumes.....

As this is a "numbers game" - where the main problem is the tiny number of samples, the only approach that makes any sense to me is to look at the national totals. Every scamera partnership will, I'm sure, be using its statisticians to try and justify its existence. They all claim wonderful reductions in KSIs (follwed by various weasely words like "at the camera sites") but nationally, the figures continue to follow their age-old trend. Basically, cameras have not lived up to their claims. Anyone remember the 30% aspirations? (AKA the 1/3 lie)! I don't think it merits any further analysis than that, to be honest. My guess is that this recession will save more lives on the road than any cameras ever did, but that's just a guess. I'm sure time will tell.

As for Greenshed's bleating on about "peer review" (not the first time anyone seeking to prop up the crumbling scamera propaganda machine has resorted to that one! ), I'm curious as to why anyone would want to have an already commonly accepted mathematical concept "peer reviewed"? Does he not believe it exists? Does he think it is wrong in some way? If I said "the total number of people killed on Cumbria's roads in 2010 was 7 more than in 2009" would he ask for a "peer review" of my subtraction? Are there some "experts" out there would dispute that 30-23=7?

There is no lesson for me on RTTM in that. I have learned something though; you seem to be confused about the requirements to install a fixed camera under the safety camera program and current DfT recommendations.
Why would a camera have been installed when one single KSI collision resulted in 4 KSI casualties? Why would a camera have been installed if that collision had resulted in, let's say, 10,000 casualties? The site wouldn't qualify.
Try doing it again with numbers that satisfy the criteria.
I see your toady friends didn't spot that error either; well done all; still confused after all of these years eh?

Let's change the argument to four single KSI collisions instead of one single KSI collision giving 4 KSIs. Is the underlying point valid?

A distinction without a (real-world) difference?

_________________
Views expressed are personal opinions and are not necessarily shared by the Safe Speed campaign

In this discussion about statistics, don't forget there are confounding factors in the analysis. Two important ones are:

Back in the mid-1990's, people were happily phoning each other up on hand held mobiles while driving. This has been much reduced in recent years.

The non-insured vehicle spotting cameras. These I suspect HAVE had a good effect on accident figures, but were brought in after the speed cameras (on average).

Why not ascribe any reduction in accidents to one or both of these factors, rather than the speed cameras?

The other scientific test is, SWINDON, before and after.

Greenshed, re "peer review":

How can anyone properly peer-review the theory when the numbers have been kept secret ? And the figures are under the control of one side of the argument?

The fact that the figures have been kept under wraps for so long tells me that it is only one side that has a lot to lose from proper independent peer review !

This particular "toady friend" was able to appreciate that the "error" wasn't in any way relevant to the argument, so he let it be. However, if you want to try and subvert this from an argument about statistical analysis, (where I appreciate you might feel uncomfortable ) to one about camera placement semantics, then well done, you spotted it, congratulations...

...now can we get back to the argument?

We're still waiting for an answer to the above Greenshed, anytime this week would be nice.

_________________
My views do not represent Safespeed but those of a driver who has driven for 39 yrs, in all conditions, at all times of the day & night on every type of road and covered well over a million miles, so knows a bit about what makes for safety on the road,what is really dangerous and needs to be observed when driving and quite frankly, the speedo is way down on my list of things to observe to negotiate Britain's roads safely, but I don't expect some fool who sits behind a desk all day to appreciate that.

And, to be fair, both GS and I are still waiting for an answer to his question as to whether RTTM can be invoked in those circumstances

_________________
When I see an adult on a bicycle, I do not despair for the future of the human race. H.G. Wells
When I see a youth in a motor car I do d.c.brown

Many of my questions prior to that still remain unanswered (three of them in one post alone; here's another [Robin's]), but I will take the high ground anyway... (not aimed at you DCB)


I'm not sure of the scenario that the question relates to.

A mean period including the baseline, being virtually zero, suggests the nominal selection criteria (history of KSI collisions) wasn't a wasn't a factor, hence no scope for RTTM.

A mean period excluding the baseline, being virtually zero, allows an elevated baseline, therefore allowing RTTM.

Either that or the question needs some clarification.

_________________
Views expressed are personal opinions and are not necessarily shared by the Safe Speed campaign

If there were three events in a ten year period prior to the implementation of some policy and only one in the subsequent ten years would you say that the improvement could be explained by RTTM?

_________________
When I see an adult on a bicycle, I do not despair for the future of the human race. H.G. Wells
When I see a youth in a motor car I do d.c.brown

My 2p worth..

Sounds like it would be better or more pertinent, as far as information gathering goes, to think of it as four events over twenty years dcb. If you narrow down the timescale or be selective over which period to analyse you could 'prove' almost anything, as in Robin's earlier nice analogy.

_________________
The views expressed in this post are personal opinions and do not necessarily represent the views of Safe Speed.
You will be branded a threat to society by going over a speed limit where it is safe to do so, and suffer the consequences of your actions in a way criminals do not, more so than someone who is a real threat to our society.

But if you are using ten year half cycles you would have to wait at least another twenty years (one more complete cycle) before you could determine if the cycles were cycling around the mean line. You can't judge anything on just one cycle.

_________________
My views do not represent Safespeed but those of a driver who has driven for 39 yrs, in all conditions, at all times of the day & night on every type of road and covered well over a million miles, so knows a bit about what makes for safety on the road,what is really dangerous and needs to be observed when driving and quite frankly, the speedo is way down on my list of things to observe to negotiate Britain's roads safely, but I don't expect some fool who sits behind a desk all day to appreciate that.