How?

[screenman]

Do they collect information on heads that were not damaged because they had a helmet on when the accident occurred, and therefore did not need to go to hospital or see a doctor.

 

[The Brewer]

Satellites

 

[benb]

They don't directly, but we can infer this.

Let's say the percentage of people wearing a helmet is 40% (I have no idea of the actual figure, this is just a wild guess)
You then look at hospital admissions, and compare the percentage of helmetted vs unhelmetted cyclists admitted for head injuries.

If helmets were effective, you would expect the percentage of unhelmetted head injuries to be greater than 40%.

 

[screenman]

So a wild guess then at best. This is where I have the problem with some statistics, only half the data needed seems collectable.

 

[Markymark]

The 'wild guess' was a stab at the figure wearing helmets, not at the rate. I'm sure that a more accurate figure could obtained. Then, as posted, the rate of helmetted v unhelmetted figures at the hospital is compared. Both of these figures can be fairly accurately measured.

That is not based on guess work, but neither is it 'proof' but it's probably a fairly good indicator.

I'm sure that this will get shot down in flames as I promised myself I will never write on a helmet thread unless in the presence of a lawyer, a doctor and a priest.

 

[benb]

As above, it's an estimate. You'll never get 100% accuracy, but methods like this are standard epidemiology, and widely accepted.

More persuasive are the countries that have introduced compulsion and not seen any improvement in the proportion of head injuries.

 

[StuartG]

BenB is wrong - or that method would result in the wrong answer. It assumes the population of helmet wearers and non-wearers is identical. But they are not. They made two different decisions for a start. The reasons are are many but one is age. I don't have the figures to hand but I'm betting the helmet ratio falls with age. However experience increases and (if we assume a pattern similar to motorists) then they will have less accidents. So if you use BenB's method this factor would show up as the helmetless having less head injuries. When in fact it is down to experience - a confounding factor.

There are many methods used to normalise populations so that the helmet factor can be inferred with a rather better level of confidence. But it is complex which is why you need professionals doing the job and checking each other. Statisticians are really good at this but rubbish at brain surgery. And they know it. The reverse can't be said for brain surgeons ...

 

[benb]

BenB is wrong - or that method would result in the wrong answer. It assumes the population of helmet wearers and non-wearers is identical. But they are not. They made two different decisions for a start. The reasons are are many but one is age. I don't have the figures to hand but I'm betting the helmet ratio falls with age. However experience increases and (if we assume a pattern similar to motorists) then they will have less accidents. So if you use BenB's method this factor would show up as the helmetless having less head injuries. When in fact it is down to experience - a confounding factor.

There are many methods used to normalise populations so that the helmet factor can be inferred with a rather better level of confidence. But it is complex which is why you need professionals doing the job and checking each other. Statisticians are really good at this but rubbish at brain surgery. And they know it. The reverse can't be said for brain surgeons ...

Sorry yes, I deliberately left out confounding variables to keep the principles simple and easy to understand.

 

[screenman]

Idiot, personally insults now, having trouble getting your own way or something.

 

[lukesdad]

Sorry yes, I deliberately left out confounding variables to keep the principles simple and easy to understand.

Never a truer word spoken when it comes to so called evidence

 

[benb]

Never a truer word spoken when it comes to so called evidence

What are you on about? I was explaining the principle by which we would try and measure the effectiveness of helmets. You obviously don't understand the meaning of the word evidence, nor the scientific method.

 

[lukesdad]

Oh I am sorry I thought you were indicating some one was simple.... Oh you are, condescending ...

 

[lukesdad]

What are you on about? I was explaining the principle by which we would try and measure the effectiveness of helmets. You obviously don't understand the meaning of the word evidence, nor the scientific method.

...and you obviously dont understand the meaning of variables. If you were truly trying to measure the effectiveness of helmets you would be taking far more details at source.

A simple one might be time of day for instance, wouldn't be hard would it seeing as its logged anyway.

any ideas why you might want to do this ?

 

[PpPete]

Collection of data @ A&E is flawed anyway. Non wearers (like me attending for an arm injury) get asked if we were wearing a helmet, and lie to save the utterly pointless lecture.

Sent from my HTC Wildfire S A510e using Tapatalk 2

 

[benb]

...and you obviously dont understand the meaning of variables. If you were truly trying to measure the effectiveness of helmets you would be taking far more details at source.

A simple one might be time of day for instance, wouldn't be hard would it seeing as its logged anyway.

any ideas why you might want to do this ?

I'm pretty sure I do understand vriables.

I used a simple example of the kind of approach you could use to measure helmets effectiveness. Of course it was incomplete - I didn't have time to design a full study. It was simply meant to illustrate why people not reporting accidents wasn't necessarily a problem.