Motorcycle Safety Foundation open to reconsidering safety study...

RAY,Wrote in Response, Motorcycle Safety Foundation open to reconsidering safety study...

As I've thought about this before I've just assumed that this study would be another opportunity for paid off investigators to turn out the same kind of trash that NHTSA turns out, failing to account even for the most obvious independent variables and report correlations without first applying standard statistical tools. Indeed, the initial thought that crossed my mind was that the powers-that-be were just looking for a study to displace the favorable Hurt findings, such as that 2/3 of all motorcycle accidents are the result of auto driver negligence without any fault on the part of the motorcyclist, etc. When the study numbers were dropped to 300, it made it obvious that it could not possibly have sufficient data to reach statistical significance, in which case a real scientist would be required by the scientific method to draw the conclusion that "the results failed to confirm the hypotheses," e.g., that wearing helmets significantly reduced the incidence of brain injury and death. At that point my thoughts on the study were mixed. If we had a good scientist involved, this would have to be the conclusion that he must draw. But on the other hand, that assumed that a good scientist would be chosen to head the study and insist on scientifically sound methodology, which was an assumption the accuracy of which I was suspect given all the crap NHTSA turns out.

But if you read this discussion below I think we first have the insight that the investigator is probably a scientist with some scientific ethics, likely both to use methodology designed to control for at least the most obvious independent variables, and his whole point is that he wants to be assured that there will be sufficient numbers to offer the opportunity to get results meaningful in science as measured by application of the conventions requiring that they report only results found to be statistically significant.

On the other hand, I think he realizes, as I did, when the number 300 was floated that it would not provide a good chance of statistically substantiating other than the most highly correlated associations. My experience is that numbers much higher than that would be required to substantiate even a "doubling" correlation, which I think would be a highly unlikely correlation, even in comparison brain injury and death resulting from brain injury in riders wearing helmets and not wearing helmets across 300 accidents. I haven't looked at the methodology and so I don't know all of the hypotheses that the study contemplates testing, and there might be others, favorable to us, as found in the Hurt study, that he would not confirm for the same reason.

I suspect what he will find is that numbers much greater than 300 will be required to provide any likelihood of obtaining statistically significant results, and so he will probably say the study can't be conducted with such small numbers. After that, they'll be back to square one, with the remaining options being either to gain additional funding for a study with adequate numbers or just scrapping the study.

Ray