HOW DANGEROUS IS CYCLING? DEPENDS ON WHICH NUMBERS YOU EMPHASISE.

On Wednesday, May 22, 2019 at 4:56:53 PM UTC-7, AMuzi wrote:
On 5/22/2019 6:43 PM, Frank Krygowski wrote:
On 5/22/2019 4:49 PM, jbeattie wrote:
On Tuesday, May 21, 2019 at 2:07:39 PM UTC-7, Frank
Krygowski wrote:
On 5/21/2019 11:29 AM, jbeattie wrote:
On Monday, May 20, 2019 at 4:10:56 PM UTC-7, Frank
Krygowski wrote:
On 5/20/2019 5:07 PM, jbeattie wrote:

Tom, statistically, you did not have any of your head
injuries. They were imagined...

IOW: "Math is HARD!!!"


It's not math. It's statistics -- where two plus two
may equal four, depending on who you are. Large
population studies say little or nothing about the risks
encountered by individual cyclists in particular areas
or engaging in specific types of cycling. Tom is an
example -- as are most of my cohorts. It doesn't take a
math genius to recognize that lumping together the
accident rates of NYC bike messengers and Sun City
retirees is going to create a combined rate that is not
accurate for either group.

Jay, that has nothing to do with your quip "Tom,
statistically, you did
not have any of your head injuries."

Obviously, that's not what the statistics say. But
unfortunately, there
are plenty of people who seriously engage in your logical
fallacy. One
way it's been expressed is "Yes, there may be only one
bike fatality per
ten million miles ridden. BUT WHAT IF THAT ONE IS _YOU_??"

What logical fallacy?

The same one that leads millions of people to waste billions
of dollars on lottery tickets. "It doesn't matter if the
odds are hundreds of millions to one against me. What if _I_
win?"

The same one that leads people to shun vaccinations for
their kids. "The scientists have numbers claiming
vaccinations don't cause autism, but what if they're wrong
about _MY_ kid?"

It's the belief that every individual is totally unique, and
that large population data can say nothing about any
person's chances of any occurrence.

Who's on the other side of this debate? Medical science, for
one - with large medicine trials that confirm that medicine
A is beneficial; and with other trials that show that
medicine B is no better than a placebo. They do this by
testing large numbers of patients; and the assumption is
that the next patient won't be miraculously different. He'll
probably respond about the same way.

Insurance companies are also on the side of statistics. They
take in billions of dollars betting against the idea that
everyone is absolutely unique. They know that there are
individual differences; but they bet heavily on aggregate
data. Of course some individuals fall far enough outside the
norm to cost the insurance folks money; but the vast
majority of their customers meet their predictions well
enough to ensure healthy profits.

Your statistics are so blunt, its like saying that a man
has a one in 1,000 chance of getting ovarian cancer
because that is the national statistic.

Of course, you have to choose the applicable data for the
proper cohort. (Although, weirdly enough, we're now in an
age where gender is purportedly a matter of opinion!)


And regarding large population studies: It's true that
every large
population has its probability distribution, usually a
bell curve. And
there are certainly individuals out on each tail end of
each bell curve
- the good end and the bad end.

But that does not mean the studies say "little or
nothing" about
individual risks. Unless the individual is riding his
bike off the roof
of a skyscraper, his individual values are best thought
of as
modifications of the mean value. One individual will very
likely be
within two standard deviations of the mean. He's very
unlikely to be
more than three standard deviations away from the mean.
Or in other
words, almost everybody is almost average.

My lifetime mileage is approaching 300,000 miles which is
a multiple of standard deviations above the norm and yet
you would put me in the same cohort as the once-a-year
beach-bike cruiser at the local resort.

Somewhere upthread, we were talking about your individual
crashes or injuries, which you proclaimed to be many.

Your lifetime mileage is extremely impressive. It would be
interesting to take your personal injury count, divide by
your lifetime mileage, and see how far you lie outside the
available averages - recognizing that the "average" data is
very rough.

Frankly, what I'd expect is that you (and most other
super-dedicated riders) would have much lower per-mile crash
rates than average. FWIW, Forester claimed this in one of
his books.

But it depends. Danny MacAskill also has tons of mileage;
but I'm sure he has tons of crashes. (He actually does ride
his bike off rooftops.) And I've known avid riders who gave
it up because they had too many crashes. Extreme risk takers
and extremely clumsy people must be a big part of the "bad"
tail of the bell curve.

Above all, if a person chooses situations and behaviors
that are well
within his skills and capabilities, he can place himself
further on the
"good" side of that bell curve. If he takes excessive
risks, he places
himself further toward the "bad" side.

An individual with a large number of crashes almost
certainly didn't get
those because statistics failed. It's because one way or
other, his
choices were bad.

Thank god you're not a doctor -- you'd ignore family
history, work exposure and every other relevant factor in
predicting whether a particular patient was at risk for a
specific disease.

All the world is not the same, and everyone in the world
is not exposed to the same risks. For example, most of
the pedestrian deaths in Portland happened on a handful of
roads. You are at risk crossing those roads -- more so
than crossing any other roads in Portland. You're crazy to
ignore the specific circumstances under which others ride,
walk, sleep, garden, etc.

I'm not ignoring them. But I'm saying almost everyone is
almost average. That's true within any properly selected
cohort.

If someone's experience falls far outside the norm for his
cohort, then something very strange is happening; or perhaps
there's been some mis-measurement.

Here's a specific example: The best data available (from
several sources) estimates that there are about ten million
miles ridden in the U.S. between bike deaths. (Actually
more, but that round number will suffice.) And the best data
I could find said about 45% of those were actually caused by
TBI. Some others claim a higher TBI percentage, although the
"75%" claim seems imaginary.

So, again using very round numbers, there are probably at
least 15 million miles ridden between bicycling TBI deaths.
Yet I've recently read a claim "My helmets saved my life
three times!"

What's the most rational conclusion? Seems to me one
possibility is that person is an ASTONISHINGLY bad rider,
way out beyond the 99.9999th percentile. Or much more
likely, that person is flat out wrong - that none of the
three head impacts would have killed him, despite his
heartfelt belief.

IOW, I don't think the people who make that claim or very
similar claims are really that far outside the norm.

And - "Completely separate issue" warning! - I think it's
still true that in most incidents when a bicyclist falls, he
(or she) made a mistake. They could have avoided it if they
had done things differently, including shunning a risk that
was outside their capability at the moment.




No wonder I feel weird I think I had 0.0000248 of a death on
my ride just this morning:

https://photius.com/rankings/2019/po...te_2019_0.html

--
Andrew Muzi
www.yellowjersey.org/
Open every day since 1 April, 1971

And that chart is a perfect example of statistical anomalies because the "low numbers of deaths" occur in rather unhealthy areas that the statistics do not have a good statistical reading on. And in general the high numbers of deaths per unit are in areas where there is a very large, dense and poor population in which medical care is almost non-existent.