Oh jeez, 15 splits….still, those that know me know I’m game!
hjeg2 wrote:
Firstly, why do you keep putting the link in? It makes it very hard to read when replying.
I didn’t. The administrators were experimenting with a keyword hotlink system, it has since been disabled. We apologise for the inconvenience.
hjeg2 wrote:
smeggy wrote:
I merely asked you to explain your stated belief [that "they are much better than nothing"] - is that not reasonable?
Of course it's reasonable, and I did.
…
You've done it again! Goodness, what is that all about? I shall quote the text of my explanation at another point as I'm in the middle of this reply.
So where is it?
I presume you agree that I did not “clearly wanted to talk about stats”.
hjeg2 wrote:
smeggy wrote:
How does peer-review exist as an issue and why?
Let's put it another way: if the data is so good then why not get it peer-reviewed? Is it not worth doing so in order to shut George Monbiot up?
You didn’t put it a first way!
Ah,
Tu quoque, the refuge for those who have no answers for questions put to them.
To answer yours (so now I have the high ground): the work involved in scientific publication is largely the framework one - establishing enough in each paper and dismissing those items of research that just don't belong. It's do-able, but significantly time consuming. Doing so will significantly divert efforts away from the campaign and momentum will be lost. In our view it is not yet worth it.
I’ll come to Monbiot later on.
hjeg2 wrote:
Please clarify.
We have "some police", but fewer on the roads than before, needlessly so because of the huge amount of paperwork they now have to do. The increased of cameras fooled people into thinking their loss was compensated; however, the continued loss of the nationwide fatality trend proves the folly of it.
Hence we can have more police on the roads without being "at the detriment of other crimes".
hjeg2 wrote:
I guess because I trust the government more. And because I am on the Safe Speed site. But yes, perhaps the government should have it's data peer-reviewed. And again, I don't accept that RTTM is proven. I have now found my source for it being wrong - I have no doubt though that you will dismiss it out of hand.
http://www.guardian.co.uk/commentisfree ... 72,00.html…
The point here is that I don't have the time to argue them. Even writing a post like this takes up a lot of my time. I'm sure you know that when you get into something like this it gets very detailed. I have found my source (or 'source' if you prefer) for my disagreement above.
That’s opinion without justification - blind prejudice.
Where were those WMDs? (weapons of mass destruction)
hjeg2 wrote:
smeggy wrote:
Then there’s the stats19 data that shows the portion of vehicles exceeding the speed limits involved in crashes (5%); that strongly supports the RTTM argument.
Again, I refer to my source above.
Your source conveniently didn’t pass comment upon the very damning stats19 reports.
hjeg2 wrote:
Er, well I do know of a study but I haven't looked at it, as explained above.
Well, had you have done you would have seen the truth instead of George Monbiot’s biased, cherry picked, irrelevant version of it. If anyone’s work should be peer-reviewed it should be his. Why? Well he gave the figure which didn’t allow for long-term trends, more bias on selection if you will (the national figure was falling due to many other measures, just as it was for decades before speed cameras were introduced) which of course also add to the illusion of perceived speed camera effectiveness.
The report he referred to had this to say about it:
The national safety camera programme, Four-year evaluation report wrote:
Since the site selection guidelines for cameras include threshold levels of both
all personal injury collisions (PICs) and fatal and serious collisions (FSCs),
it is likely that some of the observed reductions in collisions will be attributable
to regression-to-mean (RTM) effects rather than the effects of the cameras.
Whenever site selection is based on particularly high numbers of observed
collisions in a particular period of time, the sites identified will tend to be
those with more collisions than expected during the period of observation.
Such locations will then tend to have fewer collisions in a subsequent time
period (with or without a camera) simply because the collision count in the
first time period was abnormally high. This is the RTM effect. If RTM effects
are not allowed for there is a danger that the effectiveness of cameras will
be over-estimated.
…
Table H7 summarises the estimated percentage changes in FSCs attributable
to the effects of the cameras, RTM and trend relative to the observed FSCs
prior to camera installation. The overall average observed reduction in FSCs
is 55%. After allowing for trend and RTM effects, the overall average reduction
in FSCs attributable to these cameras is 10% of those observed in the
baseline period. RTM effects account for a fall of 35% with trend accounting for a further fall of 9%.
Well I never, RTTM is proven, accepted and extremely significant, and that’s excluding other effects of bias on selection – as described again immediately below:
hjeg2 wrote:
smeggy wrote:
When a speed camera is erected alongside another new road safety measure, like a pedestrian barrier/crossing, central reservation, junction re-layout, etc, or even just a danger awareness campaign, what do you think gets credited for any subsequent reduction in accident rate? Both, or just the "camera site"? Think about that – that’s sneaky huh?
No figures exist for the effectiveness of the described other measures, that in itself should be ringing alarm bells – yes?
Yes, so in the case of the A3 are you trying to claim that it was the 50 limit that resulted (I believe) in the reduction of accidents above and beyond RTTM, rather than the speed cameras?
No, you are clearly trying to put words into my mouth. I am explaining to you bias on selection. The A3 example is likely to be mostly RTTM (as explained). The limit drop may not have been significant at all; furthermore it isn’t the only example of bias on selection.
hjeg2 wrote:
smeggy wrote:
hjeg2 wrote:
Granted, the 50 limit will play a part.
"Will"? To what level? Can you substantiate that claim? (PS, even if true that’s another example of 'bias on selection').
Sorry, I thought that was something that you would have agreed with. Well if not then where's the 'bias on selection'?
No, I said it could play a part: "even if true". Are you trying to misrepresent me?
I said that in response to you saying that it "will" play a part. I asked you to justify the "will" in your statement. I presume anything of relevance still won’t be forthcoming.
Do you now understand bias on selection and that it is separate from RTTM and additional to the illusion of perceived effectiveness of cameras?
Are you not concerned that figures exist for the effectiveness of speed cameras but no such figures exist for other safety measures, even though they exist at cameras sites?
hjeg2 wrote:
smeggy wrote:
:???: Isn’t that a defining characteristic of the illusion of RTTM? If not then what’s the difference?
And again! No the point was that you're getting less accidents than simply immediately before.
Isn’t that exactly what you get with RTTM? If so then explain the difference. If not then you obviously don’t understand RTTM; read the quote of the report above (or look at my sig below for a summary)