helmets reduce head injury by 60% and brain injuries by 58%

[jules21]

despite my post above, hospitals do collect injury data that includes the cause - cycling, home fall, etc. while they aren't able to collect low level injury data, the obviously collect data on injuries requiring hospitalisation. that's still meaningful data.

[vitualis]

It seems that the people who deny the problem with helmets with the Melbourne scheme simply deny the reality of the numerous international schemes around the world.

there's a bit of sleight of hand here, where the impact of helmet laws on the melbourne bike hire scheme has been held up by opponents as indicative of the impact of helmet laws more broadly. that isn't right.

The impact on mass bike hire schemes is the most visible problem of mandatory helmet laws. With regards to its broader impact, there is some level of disagreement between the experts in the field (i.e., public health physicians). Although there isn't consensus, the bulk of published literature suggests that the public health benefits of not having mandatory helmet laws vastly outweigh the risk of increased rate of head injuries per accident. With regards to cost-effectiveness, the literature again suggests that mandatory helmet wearing as a policy gives a poor return compared to other safety policies that can be put into place.

You are correct though that the effect mandatory helmet laws have on bike hire schemes is not directly indicative of the impact more broadly.

why? surely the objective is just to assess the impact of helmet laws on ridership rates. if the impact is low and explained partly by low compliance with the law, then the impact is still low. that's a perfectly legitimate finding, to me, unless the effect of the law is different in the region you wish to apply the results to. obviously if the law isn't improving wearing rates, that's a problem in its own right.

There are two specific problems. Firstly (and most importantly) is something called statistical power. A study needs to be of a particular size / have a specific threshold for participants to be able to have the power to detect (or reject) the effect you are looking for. Although I didn't mention it in my last post, the authors of the study don't provide a power calculation so we don't know how large the effect must be for the study to be able to detect it. This raises serious questions about the statistical robustness of their findings. Assuming that a power calculation was in fact done but simply not reported, then an unexpected finding in the participant demographics (in this case, poor compliance for helmet wearing in areas with helmet laws and high rates of helmet wearing in areas without) can substantially weaken the ability of the study to make any finding of statistical significance even if a real effect exists.

Secondly, there is the issue of external validity. We don't really care so much about the specific Canadian example but rather whether those findings can be generalised elsewhere. If the study found that there was little effect of helmet laws on ridership but, this was in the setting of relatively common spontaneous helmet use prior to the law, with poor compliance to the law after introduction; the findings may not be reasonably generalised to other situations where compliance is expected to be good (or well-enforced). That is, the findings lack external validity and are not informative outside of the specific study setting.

Methodological rigour is very important. The problem is, it is not entirely obvious whether the non-detection of a statistically significant effect in the study is due to there REALLY being no effect, or whether it was due to a methodological shortcoming of the study (i.e., a type 2 error).

you're entitled to that view and you could even be right, but frankly any academic study will draw criticism, particularly from those who don't favour the conclusions. i'd be interested to see credible studies with contrary findings. their apparent absence speaks more to me than your concerns about the robustness of this study.

There are credible studies with contrary findings and as I wrote in my last post, those studies are discussed and referenced in the Canadian study. You can access and read the study through the link in my last post (I posted it on my Google Docs account). Please don't make assumptions and arguments through article abstracts.

I have some sympathy for the researchers because this type of research is extraordinarily hard to do. For example, the response rate in their surveys (which mind you the authors haven't commented on their reliability because it most likely wasn't tested --> this is also a potential source of substantial bias due to problems with memory recall) is actually pretty good.

sourcing and processing the data is a problem with many statistical studies - i understand your concerns (broadly). but again, the hypothesis is that helmet laws have reduced ridership rates. that's just not proven, even if it's become a mantra of helmet law opponents.

I agree that it is not proven but there is reasonable evidence for this assertion. The most recent Canadian study provides some possible evidence to the contrary but I believe it has substantial methodological weaknesses.

Frankly, I doubt very much that there will ever be high quality evidence one way or another. If you really wanted to study this further, the next step would be to use quasi-experimental design in a country that in planning to introduce mandatory helmet laws. Introduction of the laws can be staggered over years to different regions allowing comparison both between before and after in each region as well as the comparison between regions at different phases of the law introduction. However, governments don't introduce laws with a view to do a massive social policy experiment so this study will never take place. Ambiguities will always occur when doing retrospective studies so it is likely that the evidence base to this question is as good as it is going to get. In this setting a best guess and economic modelling should take place.

it's a problem. disposables, which were suggested in a weekend newspaper article on the topic, are unlikely to meet the mandatory Australian Standard. like much of the article, they didn't think that out very carefully..

I agree; unless the disposables were remarkably engineered which would likely make them expensive and not terribly disposable.

Someone up this thread discussed police not enforcing the rule for people using the hire bikes. I don't believe that this is a valid solution either as you cannot expect the hire operators to advise the public to knowingly break the law. A possible solution for a government unwilling to repeal the law generally might be making an exemption to the law within certain city limits to which the bicycle hire scheme is expected to operate.

Regards.

[il padrone]

you cannot expect the hire operators to advise the public to knowingly break the law.

The defacto situation for Melbourne Bikeshare currently - when faced with a system optimised to work most effectively for occasional dispersed short-term hires, and no practical access to helmets provided. Despite the operator's warning stickers and waivers, the only way this system will work well and make a financial success is for people to ignore the law.

Sad really - a chance to revolutionise and boost the use of bikes for short-term urban transport by otherwise non-riders is being missed.