Re: Helmet Poll: First Hand Experience



Espressopithecus (Java Man) wrote:
>>

> Correlation isn't causation.
>


True but when the correlation is synchronous with a step change in one
of the correlates and occurred in two separate unrelated situations at
different times you can be pretty sure there is a causal link as well
because the probability of it happening by chance becomes very small indeed.


--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham
 
Espressopithecus (Java Man) wrote:.
>>>

>> The promulgation has nothing to do with it. A step change doubling of
>> helmet wearing resulting from mandatory helmet laws produced no change
>> in head injuries amongst cyclists. When the large decrease in numbers
>> cycling caused by the MHL is taken into account head injury rates
>> actually rose on a doubling of helmet wearing. That is proof that the
>> helmets didn't work. It doesn't say why they didn't work but clearly
>> they didn't

>
> There are many possible explanations for the observed phenomenon apart
> from "wearing a helmet while cycling is less safe than not wearing a
> helmet while cycling". The data doesn't control enough variables to
> support the conclusion you've reached.
>


Most can be ruled out by a) it being a step change in helmet wearing adn
b) it happening multiple times in several countries at different times
and all correlated with the introduction of an MHL. Trust me the data
is sound even if you would wish it not to be.

--
Tony

"The best way I know of to win an argument is to start by being in the
right."
- Lord Hailsham
 
in message <[email protected]>,
Espressopithecus (Java Man) ('[email protected]') wrote:

> In article <[email protected]>, [email protected]
> says...
>> Espressopithecus (Java Man) wrote:
>> >
>> > No it isn't. It's easy to prove that the promulgation of mandatory
>> > helmet laws was sometimes correlated with increased incidence of
>> > head
>> > injuries among cyclists. That isn't proof that "helmets just don't
>> > work".

>>
>> The promulgation has nothing to do with it. A step change doubling of
>> helmet wearing resulting from mandatory helmet laws produced no change
>> in head injuries amongst cyclists. When the large decrease in numbers
>> cycling caused by the MHL is taken into account head injury rates
>> actually rose on a doubling of helmet wearing. That is proof that the
>> helmets didn't work. It doesn't say why they didn't work but clearly
>> they didn't

>
> There are many possible explanations for the observed phenomenon apart
> from "wearing a helmet while cycling is less safe than not wearing a
> helmet while cycling". The data doesn't control enough variables to
> support the conclusion you've reached.


It's entirely possible that the moon is in fact made of green cheese
except that the few small areas where astronauts have actually landed
are covered with a thin layer of mineral dust. It's possible, but it's
highly implausible. Similarly, it's entirely possible that in every
territory where MHLs have been enforced, some mysterious local shift has
occurred in the world's physics which cancels out the beneficial effect.

Occam's razor, however, tends to suggest otherwise.

--
[email protected] (Simon Brooke) http://www.jasmine.org.uk/~simon/
.::;===r==\
/ /___||___\____
//==\- ||- | /__\( MS Windows IS an operating environment.
//____\__||___|_// \|: C++ IS an object oriented programming language.
\__/ ~~~~~~~~~ \__/ Citroen 2cv6 IS a four door family saloon.
 
in message <[email protected]>,
Espressopithecus (Java Man) ('[email protected]') wrote:

>> Helmet fans can speculate, I suppose, that some of the helmets
>> prevented some head injuries.

>
> Just as easily as helmet knockers can speculate that some of the
> helmets exacerbates some head injuries...


Both of these things are undoubtedly true. What can't be doubted,
however, on the figures, is that the number in the latter case is
greater than the number in the former.

--
[email protected] (Simon Brooke) http://www.jasmine.org.uk/~simon/
;; Generally Not Used
;; Except by Middle Aged Computer Scientists
 
In article <[email protected]>, [email protected]
says...
> Espressopithecus (Java Man) wrote:
> >>

> > Correlation isn't causation.
> >

>
> True but when the correlation is synchronous with a step change in one
> of the correlates and occurred in two separate unrelated situations at
> different times you can be pretty sure there is a causal link as well
> because the probability of it happening by chance becomes very small indeed.
>

Nonsense.

In order for correlation to be rationally extrapolated to causation,
there has to be a very plausible explanation of the causal mechanism.

What is it?

Rick
 
In article <[email protected]>,
[email protected] says...
> in message <[email protected]>,
> Espressopithecus (Java Man) ('[email protected]') wrote:
>
> >> Helmet fans can speculate, I suppose, that some of the helmets
> >> prevented some head injuries.

> >
> > Just as easily as helmet knockers can speculate that some of the
> > helmets exacerbates some head injuries...

>
> Both of these things are undoubtedly true. What can't be doubted,
> however, on the figures, is that the number in the latter case is
> greater than the number in the former.
>

Correct. The pertinent question is "why"?

And no-one has posted the answer yet.

Rick
 
In article <[email protected]>, [email protected]
says...
> Espressopithecus (Java Man) wrote:.
> >>>
> >> The promulgation has nothing to do with it. A step change doubling of
> >> helmet wearing resulting from mandatory helmet laws produced no change
> >> in head injuries amongst cyclists. When the large decrease in numbers
> >> cycling caused by the MHL is taken into account head injury rates
> >> actually rose on a doubling of helmet wearing. That is proof that the
> >> helmets didn't work. It doesn't say why they didn't work but clearly
> >> they didn't

> >
> > There are many possible explanations for the observed phenomenon apart
> > from "wearing a helmet while cycling is less safe than not wearing a
> > helmet while cycling". The data doesn't control enough variables to
> > support the conclusion you've reached.
> >

>
> Most can be ruled out by a) it being a step change in helmet wearing adn
> b) it happening multiple times in several countries at different times
> and all correlated with the introduction of an MHL. Trust me the data
> is sound even if you would wish it not to be.
>

I don't doubt that the data is sound. It's your conclusion that isn't.

Rick
 
In article <[email protected]>,
[email protected] says...
> in message <[email protected]>,
> Espressopithecus (Java Man) ('[email protected]') wrote:
>
> > In article <[email protected]>, [email protected]
> > says...
> >> Espressopithecus (Java Man) wrote:
> >> >
> >> > No it isn't. It's easy to prove that the promulgation of mandatory
> >> > helmet laws was sometimes correlated with increased incidence of
> >> > head
> >> > injuries among cyclists. That isn't proof that "helmets just don't
> >> > work".
> >>
> >> The promulgation has nothing to do with it. A step change doubling of
> >> helmet wearing resulting from mandatory helmet laws produced no change
> >> in head injuries amongst cyclists. When the large decrease in numbers
> >> cycling caused by the MHL is taken into account head injury rates
> >> actually rose on a doubling of helmet wearing. That is proof that the
> >> helmets didn't work. It doesn't say why they didn't work but clearly
> >> they didn't

> >
> > There are many possible explanations for the observed phenomenon apart
> > from "wearing a helmet while cycling is less safe than not wearing a
> > helmet while cycling". The data doesn't control enough variables to
> > support the conclusion you've reached.

>
> It's entirely possible that the moon is in fact made of green cheese
> except that the few small areas where astronauts have actually landed
> are covered with a thin layer of mineral dust. It's possible, but it's
> highly implausible. Similarly, it's entirely possible that in every
> territory where MHLs have been enforced, some mysterious local shift has
> occurred in the world's physics which cancels out the beneficial effect.
>

Don't count on it.

You are apparently assuming that the underlying aggregate risk of head
injuries to cyclists was the same before and after MHLs. Yet I
understand that the studies also showed that cycling declined
significantly after MHLs. Are you theorizing that this decline was
uniform across all risk groups?

Rick
 
Espressopithecus (Java Man) wrote:
> In order for correlation to be rationally extrapolated to causation,
> there has to be a very plausible explanation of the causal mechanism.


So what is /your/ explanation for the figures?

R.
 
Java Man Espressopithecus wrote:
> Don't count on it.
>
> You are apparently assuming that the underlying aggregate risk of head
> injuries to cyclists was the same before and after MHLs. Yet I
> understand that the studies also showed that cycling declined
> significantly after MHLs. Are you theorizing that this decline was
> uniform across all risk groups?


Which is why one also corrects for the rate of other injuries, ie
non-head. It is the proportion of HI to other injuries which is
interestign because it doesn't change. If anything we can conclude that
those that most need to wear helmets already are and those who give up
cycling have a very low rate of HI. Loss of these cyclists from the
pool is directly counterbalanced by the extra protection given to those
who take up helmet wearing.

Hmm.. Need a bunch of things to match exactly, every time, in every
jurisdiction.

The null hypothesis is that helmets do nothing useful, so any changes
in population between the wearers and non-wearers do not really affect
the HI rate.

Is there a causal mechanis for this? Yes.
Biomechanical studies by King et al. show that while a helmet is good
at reducing linear accelleration within it's design range, it has
minimal effect on roatational accelleration of the brain.
A much lower degree of rotational accelleration is required to to cause
a similar amount of brain injury (from cadaver studies on human and
other primates dating back to the middle of the last century)

So we have a mechanical cause (or lack of one) that proceeds as
follows.

1. Brain injury is predominantly caused by rotational accelleration of
the brain, resulting in shearing of the connections between brain
cells.
2. Helmets are ineffective as an intervention for this measure.
3. Population studies where helmet wearing rapidly increases show no
change in head injury rates.

It is consistent and it is causal. Just what you asked for.

And it is repeatable. It has been seen everywhere there has been a
sudden step change in helmet wearing (ie a short term longitudinal
study).

And being a causal hypothesis that is supported by multiple
observations, it is predictable that the only effect of an MHL will be
a decrease in cycling, and no change in overall HI rates.

...d
 
"Espressopithecus (Java Man)" <[email protected]> wrote in message
news:[email protected]...
> In article <[email protected]>, [email protected]
> says...
>> Espressopithecus (Java Man) wrote:
>> >>
>> > Correlation isn't causation.
>> >

>>
>> True but when the correlation is synchronous with a step change in one
>> of the correlates and occurred in two separate unrelated situations at
>> different times you can be pretty sure there is a causal link as well
>> because the probability of it happening by chance becomes very small
>> indeed.
>>

> Nonsense.
>
> In order for correlation to be rationally extrapolated to causation,
> there has to be a very plausible explanation of the causal mechanism.
>
> What is it?


It doesn't need to be plausible, for that is merely in the eye of the
beholder. It doesn't have to be known either. Many have mentioned they don't
actually know for sure why the link is there, but have proposed mechanisms.

clive
 
In article <[email protected]>,
[email protected] says...
> Espressopithecus (Java Man) wrote:
> > In order for correlation to be rationally extrapolated to causation,
> > there has to be a very plausible explanation of the causal mechanism.

>
> So what is /your/ explanation for the figures?


If someone wants to prove that wearing a helmet while cycling either
increases or decreases the risk of head injuries to the wearers, they've
got some work to do. Causation has a much higher standard of proof than
simple correlation.

See: http://www.socialresearchmethods.net/kb/introval.htm

Rick
 
In article <[email protected]>,
[email protected] says...
> "Espressopithecus (Java Man)" <[email protected]> wrote in message
> news:[email protected]...
> > In article <[email protected]>, [email protected]
> > says...
> >> Espressopithecus (Java Man) wrote:
> >> >>
> >> > Correlation isn't causation.
> >> >
> >>
> >> True but when the correlation is synchronous with a step change in one
> >> of the correlates and occurred in two separate unrelated situations at
> >> different times you can be pretty sure there is a causal link as well
> >> because the probability of it happening by chance becomes very small
> >> indeed.
> >>

> > Nonsense.
> >
> > In order for correlation to be rationally extrapolated to causation,
> > there has to be a very plausible explanation of the causal mechanism.
> >
> > What is it?

>
> It doesn't need to be plausible, for that is merely in the eye of the
> beholder. It doesn't have to be known either. Many have mentioned they don't
> actually know for sure why the link is there, but have proposed mechanisms.
>

It must be plausible if you intend to establish a cause-effect
relationship using statistics.

http://www.socialresearchmethods.net/kb/introval.htm

Rick
 
"Espressopithecus (Java Man)" <[email protected]> wrote in message
news:[email protected]...

>> It doesn't need to be plausible, for that is merely in the eye of the
>> beholder. It doesn't have to be known either. Many have mentioned they
>> don't
>> actually know for sure why the link is there, but have proposed
>> mechanisms.
>>

> It must be plausible if you intend to establish a cause-effect
> relationship using statistics.
>
> http://www.socialresearchmethods.net/kb/introval.htm


Doesn't say anything about the problems of 'plausible' in this context. I
reckon risk-compensation is plausible - others don't.

clive
 
In article <[email protected]>,
[email protected] says...
> "Espressopithecus (Java Man)" <[email protected]> wrote in message
> news:[email protected]...
>
> >> It doesn't need to be plausible, for that is merely in the eye of the
> >> beholder. It doesn't have to be known either. Many have mentioned they
> >> don't
> >> actually know for sure why the link is there, but have proposed
> >> mechanisms.
> >>

> > It must be plausible if you intend to establish a cause-effect
> > relationship using statistics.
> >
> > http://www.socialresearchmethods.net/kb/introval.htm

>
> Doesn't say anything about the problems of 'plausible' in this context. I
> reckon risk-compensation is plausible - others don't.
>

So your explanation of the cause effect relationship is that people risk
adjust, and that is the reason why helmet use has been correlated with
increased incidence of head injuries in some (but not all) studies?

I'd like to see a clearer explanation of exactly what you think the
studies prove.

Rick
 
"Espressopithecus (Java Man)" <[email protected]> wrote in message
news:[email protected]...
> In article <[email protected]>,
> [email protected] says...
>> "Espressopithecus (Java Man)" <[email protected]> wrote in
>> message
>> news:[email protected]...
>>
>> >> It doesn't need to be plausible, for that is merely in the eye of the
>> >> beholder. It doesn't have to be known either. Many have mentioned they
>> >> don't
>> >> actually know for sure why the link is there, but have proposed
>> >> mechanisms.
>> >>
>> > It must be plausible if you intend to establish a cause-effect
>> > relationship using statistics.
>> >
>> > http://www.socialresearchmethods.net/kb/introval.htm

>>
>> Doesn't say anything about the problems of 'plausible' in this context. I
>> reckon risk-compensation is plausible - others don't.
>>

> So your explanation of the cause effect relationship is that people risk
> adjust, and that is the reason why helmet use has been correlated with
> increased incidence of head injuries in some (but not all) studies?


No, that's not what I said at all. I was only talking about the problems
with the word 'plausible'.

> I'd like to see a clearer explanation of exactly what you think the
> studies prove.


Which studies?

clive
 
David Martin <[email protected]> wrote:

[...]
> Is there a causal mechanis for this? Yes.
> Biomechanical studies by King et al. show that while a helmet is good
> at reducing linear accelleration within it's design range, it has
> minimal effect on roatational accelleration of the brain.
> A much lower degree of rotational accelleration is required to to cause
> a similar amount of brain injury (from cadaver studies on human and
> other primates dating back to the middle of the last century)


Don't forget the research with animals on the subject (IIRC by
DiGuiseppe et al - see the book "Accidental Injury", Springer on the
subject). While they could create all grades of brain injury with
rotational accelleration, linear accelleration only resulted in a slight
concussion or death at very high levels.

Ingo.
 
Jay Beattie wrote:

>>Attewell et al only looked at the studies which were small-group studies
>>(which have many confounders; the flaws in these studies have been
>>demonstrated previously) and completely ignored the whole-population
>>studies. "Looking only at red marbles, we find that 100% of marbles
>>are red!"

>
>
> So should we ignore the risk compensation study with kids which had 65
> or so subjects? -- Jay Beattie.


To which study are you referring?

R.
 
In article <[email protected]>,
[email protected] says...
> "Espressopithecus (Java Man)" <[email protected]> wrote in message
> news:[email protected]...
> > In article <[email protected]>,
> > [email protected] says...
> >> "Espressopithecus (Java Man)" <[email protected]> wrote in
> >> message
> >> news:[email protected]...
> >>
> >> >> It doesn't need to be plausible, for that is merely in the eye of the
> >> >> beholder. It doesn't have to be known either. Many have mentioned they
> >> >> don't
> >> >> actually know for sure why the link is there, but have proposed
> >> >> mechanisms.
> >> >>
> >> > It must be plausible if you intend to establish a cause-effect
> >> > relationship using statistics.
> >> >
> >> > http://www.socialresearchmethods.net/kb/introval.htm
> >>
> >> Doesn't say anything about the problems of 'plausible' in this context. I
> >> reckon risk-compensation is plausible - others don't.
> >>

> > So your explanation of the cause effect relationship is that people risk
> > adjust, and that is the reason why helmet use has been correlated with
> > increased incidence of head injuries in some (but not all) studies?

>
> No, that's not what I said at all. I was only talking about the problems
> with the word 'plausible'.


If you prefer another word, that's OK. My point is that establishing
causation requires more than showing a statistical correlation.

Rick
 
"Espressopithecus (Java Man)" <[email protected]> wrote in message
news:[email protected]...

> The criticism you cited seems sensible to me.
>
> But the results are still not consistent.
>
> Here's another study that contradicts your claim of consistency.
>
> http://pediatrics.aappublications.org/cgi/content/full/110/5/e60


And here's another criticism.

http://www.cyclehelmets.org/mf.html?1106

(did you see what I did there?).

> Those who say the studies "prove" that wearing a helmet while cycling
> reduces the risk of injury are, IMHO, as confused about the limitations
> of the data as those who say wearing a helmet increases the risk of
> injury. I don't think the data yet proves either proposition.


I'm pretty sure it shows that wearing a helmet doesn't reduce the risk of
injury. I'm doing quite well at finding problems with population-level
studies which demonstrate otherwise :)

cheers,
clive