Critique of a new bicycling report

On Tue, 22 Mar 2005 19:38:35 GMT in rec.bicycles.misc, Wayne Pein
wrote:

I see no reasonable reason why a 3 ft "undesignated" lane will harbor
less debris than a "designated" bike lane. In either case, motor vehicle
tire and wind blast propels debris into spaces where motor vehicles do
not operate.

and without the bogus lane stripe that creates a bicycle ghetto
where cagers will henceforth expect cyclists will always stay
whether it's safe or not, the cars will drive closer to the edge
of the road when there are no bicycles and do a much better job
of sweeping.

i've seen it over and over and over again. it works. a 14 foot
lene easily accomodates both bikes and motor vehicles at the same
time.

it's only paranoids like you that don't feel safe on the roads
that want to have cyclists ghettoized behind some useless paint
stripe. you need to learn how to ride effectively; do a google
on "effective cycling".

On Tue, 22 Mar 2005 22:56:33 GMT in rec.bicycles.misc, Wayne Pein
wrote:

Thanks Matt. Comments are always welcome as I use them to improve my
work. I've already made a few minor editorial changes (to my parent
WordPerfect document) based on other's comments, and will incorporate
any useful ones that anyone has in a forthcoming pdf version.

and you guys will all be wrong, wrong, wrong. cyclists do not
belong in ghetto lanes that do nothing to improve safety.

Dennis P. Harris wrote:

On Tue, 22 Mar 2005 22:56:33 GMT in rec.bicycles.misc, Wayne Pein
wrote:


Thanks Matt. Comments are always welcome as I use them to improve my
work. I've already made a few minor editorial changes (to my parent
WordPerfect document) based on other's comments, and will incorporate
any useful ones that anyone has in a forthcoming pdf version.


and you guys will all be wrong, wrong, wrong. cyclists do not
belong in ghetto lanes that do nothing to improve safety.



Dennis,

I think you need to read my posts and critique more carefully. I am
opposed to bike lanes.

Wayne

Dennis P. Harris wrote:

On Tue, 22 Mar 2005 19:38:35 GMT in rec.bicycles.misc, Wayne Pein
wrote:


I see no reasonable reason why a 3 ft "undesignated" lane will harbor
less debris than a "designated" bike lane. In either case, motor vehicle
tire and wind blast propels debris into spaces where motor vehicles do
not operate.


and without the bogus lane stripe that creates a bicycle ghetto
where cagers will henceforth expect cyclists will always stay
whether it's safe or not, the cars will drive closer to the edge
of the road when there are no bicycles and do a much better job
of sweeping.

i've seen it over and over and over again. it works. a 14 foot
lene easily accomodates both bikes and motor vehicles at the same
time.

it's only paranoids like you that don't feel safe on the roads
that want to have cyclists ghettoized behind some useless paint
stripe. you need to learn how to ride effectively; do a google
on "effective cycling".



Please re-read this thread and get your attributions correct. I am
opposed to bike lanes.

Regards,
Wayne

Dennis P. Harris wrote:
On Tue, 22 Mar 2005 19:38:35 GMT in rec.bicycles.misc, Wayne Pein
wrote:


I see no reasonable reason why a 3 ft "undesignated" lane will harbor
less debris than a "designated" bike lane. In either case, motor vehicle
tire and wind blast propels debris into spaces where motor vehicles do
not operate.


and without the bogus lane stripe that creates a bicycle ghetto
where cagers will henceforth expect cyclists will always stay
whether it's safe or not, the cars will drive closer to the edge
of the road when there are no bicycles and do a much better job
of sweeping.

i've seen it over and over and over again. it works. a 14 foot
lene easily accomodates both bikes and motor vehicles at the same
time.

it's only paranoids like you that don't feel safe on the roads
that want to have cyclists ghettoized behind some useless paint
stripe. you need to learn how to ride effectively; do a google
on "effective cycling".

I think you're confusing Wayne's comments, which closely parallel yours
(only in engineer-speak), and the original "study," which could have
been done in the back seat of a Ford Gargantua traveling down I-95
between adjacent exits. Only typing the study would have taken longer.

Pat

Wayne Pein wrote:

Ray Heindl wrote:

One point you
might want to address: the cyclists in the study were aware that
they were part of a study, not to mention the fact that most of
them were some sort of Sheriff's officers. This means that the
studied reactions wouldn't represent those of normal cyclists or
drivers. How likely is it that a motorist is going to buzz a
uniformed officer? (It's hard to tell from the pictures, but they
appear to be in uniform.)

Being videotaped would probably change the reactions of both
cyclists and drivers even further. But it sounds like the
investigators weren't willing to spend the time to wait for a
sufficient number of normal cyclists to pass.

Yes, these are good points. While it is ideal that subjects do not
know they are being studied, I can't envision how they would
change their behavior if they knew. Ride faster? That would be an
insignificant change. I doubt knowing they were being filmed would
alter their lateral position. Still it is worth noting.

If they knew the purpose of the study, they might well change their
lane position relative to where they would normally ride, just because
that's what the studiers were interested in. If, as should have been
the case, they didn't know the purpose of the study, the mere fact that
they were told to repeatedly ride along a given stretch of road with an
unusual white stripe might make them abnormally aware of their lane
position.

I did note
that the same subjects repeatedly rode the examination area, which
makes the data uniform.

I'm not sure that's an advantage. Each person is likely to ride a
given stretch of road in about the same place, subject to variability
due to traffic. So in effect they're really just using a smaller
sample size than they claim. 50 people doing one pass each is not the
same as 5 people doing 10 passes each. Multiple passes would be a good
way to estimate the variability in a given rider's position, though,
and would reduce the effect of the novelty of the stripe.

It seems logical that some motorists may afford more room to a
uniformed officer, assuming they were able to tell as much.

It seemed pretty obvious from the pictures, but I'm probably biased by
knowing they were some sort of officers.

Chances are most motorists did not realize they were being filmed.
Having done such work myself, I know that the cameras are
typically located unobtrusively.

One would hope so, anyway.

--
Ray Heindl
(remove the Xs to reply)

Ray Heindl wrote:


If they knew the purpose of the study, they might well change their
lane position relative to where they would normally ride, just because
that's what the studiers were interested in. If, as should have been
the case, they didn't know the purpose of the study, the mere fact that
they were told to repeatedly ride along a given stretch of road with an
unusual white stripe might make them abnormally aware of their lane
position.

They would not know the purpose of the study, but you are correct that
repeated passes could have made them hypersensitive.



I did note
that the same subjects repeatedly rode the examination area, which
makes the data uniform.


I'm not sure that's an advantage. Each person is likely to ride a
given stretch of road in about the same place, subject to variability
due to traffic. So in effect they're really just using a smaller
sample size than they claim. 50 people doing one pass each is not the
same as 5 people doing 10 passes each. Multiple passes would be a good
way to estimate the variability in a given rider's position, though,
and would reduce the effect of the novelty of the stripe.


I did not imply that uniform, homogenous data are an advantage. It is,
as you say, as if there is a smaller sample size.

Regards,
Wayne

"Wayne Pein" wrote in message
...
Matt O'Toole wrote:

Wayne Pein wrote:


"Effect of Wide Lane Conversions on Bicycle and Motor
Vehicle
Interactions" is a report produced by the University of
North Carolina
Highway Safety Research Center for the Florida Department
of Transportation. The report was prepared in cooperation
with the
State of Florida Department of Transportation, the USDOT,
and the
FHWA. It can be found online at:

http://wwwprod.dot.state.fl.us/resea.../Completed_Saf
ety.htm

In essence, 14 ft lanes were reconfigured with 11 ft
"motor vehicle
lanes" and 3 ft "undesignated bike lanes." The authors
claim the
re-striping is an improvement.

My critique is at:

http://www.humantransport.org/bicycl...ibrary/critiqu
e_conversions.pdf



Wayne, I agree with your critique.

I might have something more to say about it, after reading
both again more
carefully -- maybe later this week.

Matt O.


Thanks Matt. Comments are always welcome as I use them to
improve my
work. I've already made a few minor editorial changes (to my
parent
WordPerfect document) based on other's comments, and will
incorporate
any useful ones that anyone has in a forthcoming pdf
version.

Wayne

It's always dangerous to comment on a critique when one has
not read the original. Nevertheless:

A few minor comments.
1. I particularly liked your comments about the
higher-numbered iterations being too badly contaminated for
any comparison.
2. If you have a more direct citation for the fact that
overtaking by changing lanes isn't necessarily a hazard, that
would be helpful. You correctly note that it's up to them to
show it is a hazard, but I'm guessing they will respond with a
prima facie argument ("we don't have a citation, but it makes
sense that...")
3. I'd bury the fact that your earlier work is cited
incorrectly either farther down or in a footnote. Having it
right up front sets a nitpicky tone.
4. If I understand it correctly, riders were recruited to ride
on the lane one time as part of a study, knowing that they
were being observed. This creates considerable demand
pressure -- to behave in a way presumed to be in line with
experimental expectations (I think the psychological
literature would be Rosenthal, etc., but I'm a bit rusty on
this.) This isn't necessarily a good test of how bicycle
riders (and drivers) would respond to such signage over a
longer period of time.
A better design would have observed natural cyclist and
driver behavior over a longer period of time.

All in all, though, it's a good critique. It explains the
original study clearly enough to understand it, and makes a
number of criticisms coherently.

Wayne Pein wrote:
"Effect of Wide Lane Conversions on Bicycle and Motor Vehicle
Interactions" is a report produced by the University of North
Carolina
Highway Safety Research Center for the Florida Department
of Transportation. The report was prepared in cooperation with the
State
of Florida Department of Transportation, the USDOT, and the FHWA. It
can
be found online at:
http://wwwprod.dot.state.fl.us/resea...ted_Safety.htm

In essence, 14 ft lanes were reconfigured with 11 ft "motor vehicle
lanes" and 3 ft "undesignated bike lanes." The authors claim the
re-striping is an improvement.

My critique is at:

http://www.humantransport.org/bicycl...onversions.pdf

Wayne

I have read your critique and skimmed the original report.

I might make one additional comment: I do not understand why AVERAGE
spacing is used as an appropriate measure of distance between car and
bicycle.

1. I would not expect the distances to be normally distributed. In
particular, they are truncated at 1 inch (0 would be a collision). I
would expect this distribution to look more like a negative binomial
distribution or a truncated normal distribution -- and these have much,
much different statistical properties.

2. #1 is important because the mean will be highly affected by behavior
that has nothing to do with safety. From a bicyclist standpoint, as
long as the vehicle is 3 feet from the cyclists, whether it is 3 feet
or 13 feet away is of no consequence. What the cyclist really cares
about is being buzzed at very close distances.

3. Some strategies to deal with #1 and #2 might include
(a) evaluating the data is terms of "too close" versus "far enough",
similar to the analysis done in terms of whether the motor vehicle
moved into the other lane or not.
(b) truncating all observations of distance from the cyclist at, say,
4 feet. If the car is 10 feet away, count it as 4 feet away. Another
possible alternative would be to take the log of distance, but the
audience probably doesn't think in log space so this would make the
article more confusing to its intended audience.
Truncating would not affect the validity of the F-test done on
the data. While truncating does make normal data less normal, this
data isn't normal to begin with.

Bob Dole wrote:


I have read your critique and skimmed the original report.



Bob,

Thanks for the thoughts. My comments are below.

I might make one additional comment: I do not understand why AVERAGE
spacing is used as an appropriate measure of distance between car and
bicycle.

1. I would not expect the distances to be normally distributed. In
particular, they are truncated at 1 inch (0 would be a collision). I
would expect this distribution to look more like a negative binomial
distribution or a truncated normal distribution -- and these have much,
much different statistical properties.


I'm not a statistician, but from a Google search I don't believe a
negative binomial distribution is applicable for
this.

It is unclear whether the distributions of passing distances are
differently truncated for the different before and after conditions of
no-stripe and stripe because the authors did not discuss this.

2. #1 is important because the mean will be highly affected by behavior
that has nothing to do with safety. From a bicyclist standpoint, as
long as the vehicle is 3 feet from the cyclists, whether it is 3 feet
or 13 feet away is of no consequence. What the cyclist really cares
about is being buzzed at very close distances.


One could also argue that as long as the vehicle is 1 inch away that is
OK since the bicyclist didn't actually get hit, which is what the
cyclist really cares about. That is how the authors seem to treat
bicyclists.

3. Some strategies to deal with #1 and #2 might include
(a) evaluating the data is terms of "too close" versus "far enough",
similar to the analysis done in terms of whether the motor vehicle
moved into the other lane or not.
(b) truncating all observations of distance from the cyclist at, say,
4 feet. If the car is 10 feet away, count it as 4 feet away. Another
possible alternative would be to take the log of distance, but the
audience probably doesn't think in log space so this would make the
article more confusing to its intended audience.
Truncating would not affect the validity of the F-test done on
the data. While truncating does make normal data less normal, this
data isn't normal to begin with.




I think truncating as you suggest would be have been a reasonable
approach for the authors.

Regards,
Wayne