On Sat, 31 Mar 2018 00:48:29 -0700 (PDT), Andre Jute
wrote:

On Saturday, March 31, 2018 at 6:04:48 AM UTC+1, Jeff Liebermann wrote:
My guess(tm) is that the reason
there was no obvious change in the accident rate was because the
number of bicycle accidents was sufficiently small and subject to
radical variations in number, that any change precipitated by safety
clothing would disappear in the noise.

Possible.

Hard to tell, but I don't want to burn $30.50 for the report to find
out.

Also possible that, for entirely random reasons, the number of bicycle
accidents could be within a very narrow range over a quite substantial
period.

Yep. My apologies for the topic drift, but I spent some time dealing
with a similar effect when attempting to correlate the effects of cell
phone RF exposure with brain cancer. Cell phone use increased
dramatically starting in about 1995 and continues to increase today.
One might expect there to be a noticeable increase in the incidence of
new brain cancer admissions to hospitals if that were the case.

"Brain cancer incidence in SEER 9 areas of US"
https://seer.cancer.gov/faststats/selections.php?run=runit&output=1&data=1&statistic =1&year=201701&race=1&sex=1&age=1&series=cancer&ca ncer=76
Hmmm... no dramatic increase since 1995. The slight peak and decrease
is caused by the introduction of PET (positron emission tomography) to
diagnose brain cancers much earlier than before, which had the side
effect of increasing the brain cancer rate. After a while, PET scans
became the norm, the curve flattened, and the incidence rate returned
to its normal level pre-cell phone levels.

So it should be with bicycle accidents. If effective, a large number
of riders switching to high visibility clothing should produce a
corresponding decrease in accident rate. The key here is the "large
number of riders". If the statistical population sample were large, a
corresponding decrease in accidents might be considered valid.
However, if the number of riders involved were small, which implies a
rather jagged and widely varying graph of accidents vs time, then any
changes produced by a change of clothing reflectivity would be lost in
these variations (i.e. lost in the noise).

An example is nearer home to you than Italy: A few years ago, when
I explained to Franki-boy that cycling in the States is actually
much safer than he claimed, because he'd done the statistics
incompetently, I discovered that annual bicyclist fatalities
numbered for years on end in a rather narrow range around, if
memory serves, around 700. The trendline was essentially flat,
bearing no relationship to the growth in bicycles.

Yep, very much like the cell phone to brain cancer graph. According
to this site:
https://www.statista.com/statistics/227415/number-of-cyclists-and-bike-riders-usa/
there are 66 million cyclists in the USA. 700 accidents is a tiny
percentage of the bicycle riders who are eligible to becoming a
statistic (0.001%). That makes any accident survey susceptible to
huge distortions from coincidental sources, such as season, weather,
road construction, emergency medical availability, riding habits, etc.
My guess(tm) is to establish a minimum test sample of cyclists, I
would need to issue standardized reflective clothes to at least 7,000
cyclists (10%), rigorously control their use, and limit external
factors. For example, reflective clothes lose much of their
effectiveness when filthy. Issuing a reflective vest to a mountain
bike rider in winter is guaranteed to produce a dirty vest. So, 7,000
riders would be required to wash their reflective vest after every
ride. Ummm... I don't think that will work very well as most people
would simply lie and not wash the vest.

In effect, even with large numbers of novice cyclists coming
into the numbers every year, one had to conclude that cycling
was nonetheless getting to be safer; next you would have to
conclude that dedicated cycle-facilities were actually working,
that night was day, and other patent foolishness. The kicker
is that the numbers that caused me to perform a double-flip
were actually the best available government numbers.

Garbage in, garbage out. However, when obviously deficient statistics
are the only numbers available, one has to make do with what is
available. I'll take marginal numbers to bad logic, assumptions, and
guesswork any day.

Did you know that the number of bicyclists killed in collisions with
stationary objects correlates well with the number of ABA (american
bar association) lawyers?
http://tylervigen.com/view_correlation?id=6141
and the rainfall in California:
http://tylervigen.com/view_correlation?id=1490

The fallacy also works for the absence of evidence. (Absence of
evidence is not evidence of absence). An unchanged accident rate
after the introduction of mandatory reflective clothing does not mean
that reflective clothing does NOT have an effect on accident rate.
There could easily be a counter balancing effect. For example, it
might be that riders tend to ride more aggressively when wearing a
reflective vest on the assumption that the vest would protect them
from harm. At the same time, vehicle drivers would more easily notice
bicyclists. The two effects cancel each other resulting in an
unchanged accident rate.

I have no great expectation of this Italian study proving anything
more than that academics want to publish papers.

"More research and funding are necessary."
All research papers end like that.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558