It's an interesting idea, but the problem is how scientific research and statistics is done. Statistically significant means that the probability of your observed difference being due to chance is less than or equal to five percent if the null hypothesis were true. So, in this case the null hypothesis is that helmets do not help at all. You collect data from two groups helmet wearers and non-helmet wearers and see if anyone has a higher incidence of brain damage etc.. The problem is that the incidence of brain damage in any bicycling accident is really low. I've been in like about 4 myself and I've never hurt my head. So all this data is going into the against bike helmet side of the statistical hypothesis. But this testing situation is not what we are interested in. We are interested in when you get in an accident and your head contacts something do helmet decrease probability of brain damage. So, if you really want to test this hypothesis you need to look at these data only. And physics and information about what the human brain can sustain tell us that the impact of a bike crash where the head hits something is going to do damage to the brain (i.e. the brain will get bruised). So Sir Karl Popper, the godfather of scientific reasoning (in the modern age) tells us that we cannot base a science on null results. Null results are finding a lack of statisticial significance (i.e. bike helmets are not significantly different than not bike helmets). Because null results are ambiguous and can me 1 of two things: there really is no effect (bike helmets don't help) or the research was conducted poorly (i.e. they aren't looking at the right data), we can never really say anything about null results (because there is no way to know why they got null results). I bet if you looked at only cases that mattered, crashes where the cyclist hits their head, you would see a large significant decrease in brain injury. But lets go back to the p < .05 criteria for significance. For the safety of cyclists we don't want the test to be conservative against the safer thing to do, but these tests are set up to make the null hypothesis hard to disconfirm. That is we want to be really really sure that the null hypothesis is false (bike helmets do no good) before we actually go out and say scientifically that this is so. If you look at the history of lead in gasoline, the same arguments were made. There was no proof that removing lead made things safer for kids (using p < .05 criteria). But we now know that lead has very detrimental effects on kids health. The problem was with how the hypothesis was set up. Conservative in favor of the harmful thing, they're working on ways to figure out how to run better tests but basically we should probably approach the question in reverse..do bike helmets help, if we disprove this null hypothesis, then we should move on, but you won't disprove this hypothesis. . I hope this makes sense scientifically. I'm a psych stats TA at UW so, I know a little bit about stats, but not everything.