Originally Posted by
cyrano138
Here's where we're misunderstanding each other. It is plausible, even likely given the alternative explanations, that bicycle crashes don't result, by and large, in serious head injuries to the extent that using helmets will reduce the number of deaths from head injuries among cyclists.
The study itself and the results of the testing situation you're interested in are not mutually exclusive. It's possible that helmets, in the case where the head contacts something, are tremendously helpful in reducing injury to the head, but that the head rarely connects with things, during an accident, in a way that makes this important.
I know this is where people will come back with floods of anecdotal evidence about how a helmet saved their lives, but that's not the point. The point is that the study suggests that if you compare these people to the vast numbers of people riding bikes with and without helmets, they will be a very, very small group, and not enough to warrant mandatory helmet laws. This is the same style of argument I hear all the time from people who don't worry about the negative effects of smoking because their grandma's uncle smoked three packs a day until he was ninety, and then died of liver failure.
By the way, when I say, "statistically significant," I'm doing a little bit of hand-waving. Thanks for the clarification. I'm two classes away from a BA in mathematics and a BA in philosophy, but have been mercifully exempt from taking any stats classes.
I wasn't responding to you, (truthfully I just went right to the article and these are the obvious flaws with their studies that are leading them to make poor conclusions).
"It's possible that helmets, in the case where the head contacts something, are tremendously helpful in reducing injury to the head, but that the head rarely connects with things, during an accident, in a way that makes this important."
No it doesn't, science is driven by theory and if your theory is that helmets don't do anything when crashes don't involve the head, this becomes a unimportant question and an unimportant argument because helmets are only designed to help when a crash involves the head. By setting up the testing situation in this way you are not answering any important question, you are simply stating the obvious helmets don't help when the crash does not involve the head. Since the majority of crashes don't involve the head including all the crashes that don't involve the head leave you with low statistical power (the ability to see a significant difference if one exists).
The second problem was why is the null set as helmets don't work. In science you can never prove anything. But you can disprove things. And you disprove things very conservatively. By setting the null to helmets don't work you are tipping the scales in that direction. But you can only disprove this null hypothesis with a significant result. If you find a non-significant difference you haven't proven the null (because you can never prove anything). In fact Sir Karl Popper (the father of modern scientific interpretation) says that you should not and cannot interpret null results (which is what the authors of these studies are trying to do) because they can mean that A) helmets don't work (which is what you're suggesting) or B) the methods of the study are bad (which is what I am suggesting). However, I win because the conclusions are being based on null results, which means there is no proof of anything (google Karl Popper and do some homework). And we haven't dis-proven your null that helmets are innefective (thus helmets would be good) but we also haven't proven that helmets are ineffective either. In effect these studies say nothing. Seriously, this is why null results should not be published. Just because they get published doesn't change how they can be interpreted (A OR B).
The second thing is that this is a public safety question and classic philosophers say that public safety and lives should not be gambeled with, thus we should do the safe thing (wear a helmet) until data disproves that helmets are effective. Unfortunately these studies do not do that, and this is why null results shouldn't get published because people without a background in the theory behind scientific stats misinterpret and over extend the findings.
Fact of the matter is these null results (lack of a difference between the two groups) do not lead to any information whatsoever. Because null results are uninterpretable. This is a null result, thus we have no new information on the original question.
In addition, I'm saying this was the poorest science ever conducted, the question should not be do helmets help in all types of accidents. They simply weren't intended to help in all kinds of accidents and the information on the helmet box itself says this. However, the null results are still null results and SHOULD NOT BE Interpreted (seriously ask any statistics teacher).
http://en.wikipedia.org/wiki/Karl_Popper
So before you make dumb conclusions based on bad science, I want you to know that you are overstretching the data.
http://en.wikipedia.org/wiki/Null_result
Think about it this way Einstein destroyed Newton's theory of gravity with one simple observation (light bends around the sun...i.e. space is not a plane but more gooplike and bendable). But his observation was significantly different from that expected. Here the results are not significantly different and thus are "OF NO CONSEQUENCE".
Sorry to be a DB about this, but your misinterpreation of science is giving it a bad name.