bike accident fatalities and google maps?

On Sep 21, 11:51 pm, damyth wrote:
A recent spate of bike fatalities in the local area (Sand Hill Road
near Menlo Park/SLAC, CA) has me wondering. Is there a map or
database pinpointing locations of bike fatalities? This would be
incredibly helpful for people to find out where the most dangerous
roads to ride are, a geographical representation of frequency, and
maybe help people to avoid/improve those areas.

As the discussion has showed, it's really unlikely you'll ever get
such a map. There are far too few bike fatalities to show anything
significant. It would be like asking for a map of meteorite strikes,
so you could carry a metal umbrella when you were in the area.

But about the "spate of bike fatalities": While some areas are
certainly more dangerous than others, it's probable that many, if not
most such "spates" are just random variation. IIRC, Portland Oregon
had such a spate in 2006, with (I think) five cyclist fatalities that
year. That doesn't mean Portland suddenly got very dangerous. In all
probability, it just means random chance played out that way.

One interesting consequence is that, in response to a spate of
accidents, the public often calls for officials to _do_ something.
And the officials often comply, to stay elected. And the success is
often immediate - as in, "Look, we made cyclists wear pink shirts, and
the fatalities dropped back to their previous value! Aren't we wise!"

But it's probable that the observed effect is nothing more than what's
called "regression to the mean." Loosely speaking, that means
meteorites don't strike twice in the same place.* If nothing at all
had been done, the odd coincidence that caused the blip would have
vanished on its own.

IIRC, the book _Risk_ by John Adams discusses many examples of this
phenomenon, regarding improvements to "black spots" on British roads.
He claimed evidence of a strong "regression to the mean" effect.

My approach would be: if there's one particular road that
_consistently_ generates injuries or fatalities, then yes, there's
probably something wrong there. Working to fix it or avoid it might
make sense. But most fatalities are going to be almost as random as
meteorite strikes - except, as Carl noted, for the influence of
alcohol.

So I'd say, don't drink and ride. I'd also ride competently, because
many car-bike crashes are apparently the cyclist's fault. Those will
markedly improve your already excellent odds.

But I don't think it's worth the time to search out specific fatality
data.


*Unfortunately for the proverb writers, lightning often does strike
twice in the same place.

- Frank Krygowski

On 2007-09-22, wrote:
On Sat, 22 Sep 2007 13:01:05 -0400, Matt O'Toole
wrote:

On Fri, 21 Sep 2007 20:51:22 -0700, damyth wrote:

A recent spate of bike fatalities in the local area (Sand Hill Road
near Menlo Park/SLAC, CA) has me wondering. Is there a map or
database pinpointing locations of bike fatalities? This would be
incredibly helpful for people to find out where the most dangerous
roads to ride are, a geographical representation of frequency, and
maybe help people to avoid/improve those areas.

This is the kind of thing the Bikeleague could be doing, but they're not
even smart enough to have a news blog with an RSS feed, and all the neato
possibilities with that. Email newsletters are *so* 12 years ago.

They're advertising for new board members.

Use this database, select a state, scroll down to pedalcyclist
fatalities, note the statewide numbers, click on it, and look at the
county-level maps:

http://www-nrd.nhtsa.dot.gov/departm....htm#CTY_MAPS9

That's California, probably the best-represented state on RBT. The
figures and the maps show why there's so little interest in tracking
bike deaths locally. Los Angeles County, for example, had 24
pedalcyclist fatalities in 2006:

The county the location he's asking about is in is San Mateo, which must
have a population of close to 1 million and yet had 1 fatality in all of
2006. The 2 (or more? I only heard about 2) deaths on that one stretch of
road this year stand out in comparison, but the numbers are so small that
this seems like nothing more than random variation.

That particular road is fairly heavily travelled by bicycles since it
is part of one of the nicer lunchtime loops to ride as well as leading
to a couple of the better climbs into the coastal mountains. I don't
think the fact that a couple of cyclists got hit there says it is
necessarily dangerous, it just has a lot of cyclists to hit.

Dennis Ferguson

On 2007-09-22, Wayne Pein wrote:
damyth wrote:

A recent spate of bike fatalities in the local area (Sand Hill Road
near Menlo Park/SLAC, CA) has me wondering. Is there a map or
database pinpointing locations of bike fatalities? This would be
incredibly helpful for people to find out where the most dangerous
roads to ride are, a geographical representation of frequency, and
maybe help people to avoid/improve those areas.

Some of you might point out it should be "obvious" where the most
dangerous areas are, but I ride that area often, but it never occurred
to me that so many fatalities have occurred there until someone
recently pointed it out.


How do you know the bicyclists weren't at fault, meaning the roads are
not dangerous but dangerous riding resulted in the fatalaties?

With respect to the two deaths I know about there this year, both
were experienced cyclists, riding during the day, who got hit by a
car from behind. One got hit on the shoulder (and the shoulder is
10 feet wide there, with good asphalt and usually clean) by an old
guy who came over the fog line to get him, the other I think got hit
on or near a freeway overpass by a driver who had an open container of
alcohol in the truck.

I don't think the cyclists did anything wrong. I also don't think
that road is particularly dangerous, though it seems a few dangerous
drivers use it.

Dennis Ferguson

http://biz.yahoo.com/brn/070914/2258...pf=family-home

I invite all of you (including all you naysayers) to join me in
mapping out bicycle fatalities he
http://www.wikyblog.com/map/damyth/SFBayBikeFatal

_Anyone_ can make edits to this map. This is a wiki and google map
rolled in one. I started out with the San Francisco Bay Area because
this is where I happen to ride most, but there is no technical reason
why this can't include all contiguous 48 states.

It is well known that geographic information system (GIS) tools can
improve bicycle and pedestrian safety. Just google "GIS bicycle
safety." However, one should not make any hasty judgments regarding
the data.

On Sat, 22 Sep 2007 23:51:33 -0700, damyth
wrote:

I invite all of you (including all you naysayers) to join me in
mapping out bicycle fatalities he
http://www.wikyblog.com/map/damyth/SFBayBikeFatal

_Anyone_ can make edits to this map. This is a wiki and google map
rolled in one. I started out with the San Francisco Bay Area because
this is where I happen to ride most, but there is no technical reason
why this can't include all contiguous 48 states.

It is well known that geographic information system (GIS) tools can
improve bicycle and pedestrian safety. Just google "GIS bicycle
safety." However, one should not make any hasty judgments regarding
the data.

Dear D,

It will be hard to make hasty judgements, given the scarcity of bike
fatalities versus the geographical spread.

Expect a single pedalcyclist fatality per year for San Francisco
County--there's been 1 every year from 2002 to 2006:

http://www-nrd.nhtsa.dot.gov/departm...ounty_2006.HTM

Expect 0, or 1-5 deaths per year in all but one nearby county, judging
by this map:

http://www-nrd.nhtsa.dot.gov/departm....htm#CTY_MAPS9

For "safety", head north. The map shows that the ten north-most
counties in California had 0 bike fatalities in 2006.

Florida appears to be far more "dangerous," with half as many people
as California, and the same number of deaths from 2002 to 2006.

Cheers,

Carl Fogel

On Sep 23, 2:51 am, damyth wrote:
I invite all of you (including all you naysayers) to join me in
mapping out bicycle fatalities hehttp://www.wikyblog.com/map/damyth/SFBayBikeFatal

_Anyone_ can make edits to this map. This is a wiki and google map
rolled in one. I started out with the San Francisco Bay Area because
this is where I happen to ride most, but there is no technical reason
why this can't include all contiguous 48 states.

It is well known that geographic information system (GIS) tools can
improve bicycle and pedestrian safety. Just google "GIS bicycle
safety." However, one should not make any hasty judgments regarding
the data.

hear's the audience:

http://www.google.com/search?q=San+f...e7&rlz=1I7GGLJ

On Sep 23, 2:51 am, damyth wrote:
I invite all of you (including all you naysayers) to join me in
mapping out bicycle fatalities hehttp://www.wikyblog.com/map/damyth/SFBayBikeFatal

_Anyone_ can make edits to this map. This is a wiki and google map
rolled in one. I started out with the San Francisco Bay Area because
this is where I happen to ride most, but there is no technical reason
why this can't include all contiguous 48 states.

It is well known that geographic information system (GIS) tools can
improve bicycle and pedestrian safety. Just google "GIS bicycle
safety." However, one should not make any hasty judgments regarding
the data.

I am pretty curious what you'll do with the information you get. Will
you stay off a road if a cyclist had a heart attack while riding?
(That's two of your fatalities.) What if the fatality was caused by a
kid riding directly off a sidewalk into a turning truck? (That was
the last fatality in our area.)

If, say, a two mile stretch of road has two fatalities in ten years,
was that because the road is dangerous, or because of a combination of
random bad luck and a very popular cycling route?

And if you're at the east end of that two miles, and you need to get
to the west end, how much of a mountain will you climb to avoid it?

It's going to be interesting to see how - and if - this progresses.
Keep us posted.

- Frank Krygowski

crunching numbers toward a reality is difficult. Cal's two number sets
with shifting county colors is ....
but North Cal is easy. No one lives in North Cal, roads are dangerous,
and there's no traffic, and no bicycles = relatively few bicycle
fatalities.
Like the Indian mortality in Manhattan 1865-1890.
From a distance, we may comfortably ignore the North Cal data and
proceed on topo, road map, and federal stats (whom by the way became
the country's majority employeer)
Butbutbut in SF, we see a cluster. Why? well, we don't know but the
cluster is evidence that the area is relatively dangerous: otherwise
no cluster. Adding the cluster to topo, road map, and federal stats
and there's whatchucallur hard data.
Not only for the wannabe x country or x nurb rider but also the civic
do gooder.
Off course, if one needs military service to learn how to shower...?

On Sep 23, 7:38 am, wrote:
On Sep 23, 2:51 am, damyth wrote:

I invite all of you (including all you naysayers) to join me in
mapping out bicycle fatalities hehttp://www.wikyblog.com/map/damyth/SFBayBikeFatal

_Anyone_ can make edits to this map. This is a wiki and google map
rolled in one. I started out with the San Francisco Bay Area because
this is where I happen to ride most, but there is no technical reason
why this can't include all contiguous 48 states.

It is well known that geographic information system (GIS) tools can
improve bicycle and pedestrian safety. Just google "GIS bicycle
safety." However, one should not make any hasty judgments regarding
the data.

I am pretty curious what you'll do with the information you get. Will
you stay off a road if a cyclist had a heart attack while riding?
(That's two of your fatalities.) What if the fatality was caused by a
kid riding directly off a sidewalk into a turning truck? (That was
the last fatality in our area.)

You'll notice I have explicitly refrained from drawing conclusions.
It's not as if I've tried to obscure the cause of death (in the
possible heart attack cases).

If you've ridden the area at all, you'd understand why heart attacks
may have occurred in those instances. They are STEEP, and at a very
minimum the elderly should not ride those roads without having a
proper medical to verify their tickers won't give out.

I've even included incidents such as a bicyclist who crossed train
tracks, and 8 year old kid drivers. Whether those data points should
be included or not, that's exactly why technology is useful. They
viewer chooses which data set is of interest.

If, say, a two mile stretch of road has two fatalities in ten years,
was that because the road is dangerous, or because of a combination of
random bad luck and a very popular cycling route?

And if you're at the east end of that two miles, and you need to get
to the west end, how much of a mountain will you climb to avoid it?

I'm well aware of the pitfalls of cluster analysis.

Sand Hill Road has had three fatalities with the last 8 years, and an
additional one that could easily have been fatal. 3 of the 4
incidents involved female bicyclists. This is exactly the kind of
relationship one wants to elicit from this type of representation, and
a little data mining.

I've always thought of Sand Hill Road as not particularly dangerous,
although I consider it to be a "fast" road. Based on the information
I've gathered, you bet I'm going to be more situationally aware while
riding Sand Hill Road. Does it mean I'm going to stop or curtail
riding it? Of course not.

It's going to be interesting to see how - and if - this progresses.
Keep us posted.

Yeah, keep on snickering. The Google mashup presents the data from a
different perspective, it's really no different from a phase space
representation. People who mis-interpret the data or draw incorrect
conclusions are not my problem. At a very minimum, this will inform
bicyclists and drivers to be a bit more heedful of each other on Sand
Hill Road.

As Thomas Paine once said, "Lead, follow, or get out of my way."

- Frank Krygowski