On 20 Apr, 11:41, Peter Clinch wrote:
Toom Tabard wrote:
One has, however, to also be aware that when there seems good
empirical reason for a public health or safety initiative, its
introduction is accompanied by an initiative to collect full and
correctly classified data to measure the effect. This data is then
frequently compared to the incomplete and inaccurate data from before
the initiative and can result in considerable disparity between the
statistical result and the expected effect. That frequently masks the
close correspondence between the expected and actual effects.

So you start with data that's apparently good enough to act as a "good
empirical reason for a public health or safety initiative", but it turns
out it's so bad it will allow a doubling of the wearing rate in a very
short space of time to make no impact on serious head injury rates when
you look at the data afterwards?

And it turns out it's magically just as bad everywhere you look at the
population level, reproducibly so.

And it also turns out when you haven't had such a law and consequently a
big change in the data collection methods, and have a good hard look at
the statistical record in light of naturally evolving wearing rates,
that there appears to be no effect on serious head injuries at the
population level as wearing rates change naturally.

And it turns out where disparate groups (for example, UK juvenile males
and females) have different wearing cultures, their serious injury rates
aren't appreciably differentiated.


No I don't think I said, implied or accept any of that. And I won't
even attempt reasoned discussion of your recognisably specious
twaddle.

Toom