Archive for the ‘Science and Research’ Category

Crash Data Fallacies

July 12, 2013

LTRs know that I have been working with crash data for quite some time now.

I have gone from being confused about the cause of crashes to having a good idea on how crashes occur and back to being confused. I think confused is the right place to be.

To recap, there are only a few parameters which determine whether you will live or die in a bicycle crash.

First is speed, how fast you are going and how fast the cars are going. It you stay below twenty miles per hour, you’ll probably live. If you ride in areas where the average traffic speed is twenty five or below, you’ll probably live.

That’s it. All the other data is gibberish because it does not predict future crash survival.

Deadly collisions occur everywhere and they kill people regardless of skill, lane positioning, and helmets.

Part of the problem, I think is that this is NOT what people want to hear. They want to ride fast AND feel safe.

Because of this, we get the notions of Crash Data Analysis. The idea is to look at crash data in order to determine why and how the crash occurred so you can, hopefully, learn lessons that you can use to prevent yourself from getting into a crash.

On the face of it, this is a good idea. However, there are many problems with it. One of them is that the biggest proponants of this, routinely obsess over smaller dangers while ignoring larger ones.

For example, what’s more likely? A right hook or a left one? What happens more often collisions in the intersections or in the travel lane? What’s more likely to kill you, a dooring or a rear ending?

The answers to these questions depend on how you lump the data together as well as what sources you look at. You can make “edge riding” look more deadly by combing all kinds together, legal (bike lane) and illegal (sidewalk riding). Or you can make “taking the lane” seem safer by cutting the data into sub-catagories based on time and fashion choices while concealing the total.

If you look at the data from the point of view of deciding where and how to ride, you’ll find that you matter where you choose to ride, you can find people who died riding there. This does not make all cycling unsafe but it does tell us that there are limitations to crash data analysis.

Malcolm Gladwell is guilty of making a similar mistake. [http://askakorean.blogspot.com/2013/07/culturalism-gladwell-and-airplane.html]

“Yet, like a golf tournament filled with the world’s greatest players, air travel is a marvelous display of perfection: airplanes manage to land millions of times every year with very few accidents.”

The same is true for cycling. Most cyclists, whether they ride on the sidewalk, with or without a helmet, inside or outside the door zone, on a path, on a track, on in the middle of the road do so safely.

If you think about all the dangers, the lack of cycling education, the “conflict points”, the negligent driving, the rapant law breaking (by motorists and cyclists), and the lack of focus (mostly by motorists, but not always), you will wonder how someone can ride a bicycle anywhere safely. But they do. More often than not. A LOT more often than not. And most of us don’t wear helmets, don’t take classes, and don’t like to ride next to high speed cars. But we’re still more likely than not to survive a given bicycle ride.

“Like the golf gallery surrounding an errant ball landed among the trees, we surround and gawk at every minute detail of the latest airplane crash. We run through all kinds of scenarios about what went wrong, and talk about them.”

“I am not an exception; watching a tournament, I also fixate on the golfers’ mistakes. When I see a golfer hitting a poor shot, I take a moment trying to recreate the swing in my mind, trying to see if I could identify what went wrong. I picture the golfer making his approach to the ball; the stance; the back swing; the alignment of the club head when the back swing reaches the top; the down swing; location of the hip during the down swing; the follow-through. Then I think about the path of the ball flight, and try to identify which part of the swing contributed to the deviation from the intended path.”

There is nothing wrong with doing this. But we are only doing so AFTER the fact. The worst part is that we assume that what they were doing: riding with ear phones, sidewalk riding, etc, contributed to the accident. But doing so ignores all those thousands of times when we do those things and we are safe. It also ignores all those times when people do things the so-called RIGHT way, and they still die. This happens far, far more often than people think. And all the safety organizations in the world, who teach behavior modification, can not save a single cyclist from a motorist who’s out of control. There’s no place on the road or even the sidewalk that will make you safe 100% of the time.

But there are ways to make the roads safer. These methods are tested, reliable, and ought to be implemented.

Introducing Haddon: Vehicular Cycling and the Book of Job

June 24, 2013

Hands down, one of the best safety researchers ever, is William Haddon MD.

Good stuff. Let’s look at one of his papers: “Energy Damage and 10 Countermeasure Strategies”

[http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1067540/]

Today, these finds are commonly know as the Haddon Matrix. [http://en.wikipedia.org/wiki/Haddon_Matrix]

“The Haddon Matrix is the most commonly used paradigm in the injury prevention field.”

I’ll just summarize this but suffice it to say to call those in danger “stupid” or “uneducated” is not on his list. It mainly talks about controlling the agent (in cycling the “agent of death” is most likely cars) and controlling the interaction between the agent and the host (aka dedicated cycling infrastructure).

He also speaks of the Book of Job and divine punishment. It seems that Vehicular Cyclists believe in this kind of thing: “The suffering of oneself or some group is divine and well deserved punishment. Therefore sins can be expiated by appropriate change in behavior, it may be ‘too bad’, but there is nothing else to be done to ameliorate the personally or societally undersirable happening unless its an increase in efforts at human reform.”

Think about this next time you here Keri, from Savvy Cycling talking about changes in “beliefs” or “behavior”. Or when we hear about “scofflaw” or “uneducated” cyclists who “deserved what they got.”

Perhaps there’s more that can be done to make the future safer than to hope for the Great Pumpkin of an Educated Populace. Haddon thought so and he’s credited for saving over 5,000 a lives a year by changing things besides behavior, which he thought was of secondary importance, mainly because it does not work.

Bike Boxes: A Big Success

June 18, 2013

Wonderful news:

http://www.portlandmercury.com/images/blogimages/2012/10/16/1350403953-bike_box_letter_merc.pdf

The article notes that they installed 11 bike boxes in Oregon. In most locations crashes were reduced slightly or roughly stayed the same.

In 4 sites, the number of crashes seems to go up. However, all four higher crash sites are down hill. LTRS know that downhill cycling is twice as dangerous as flat ground cycling.

So it was predictable that these spots would be the trouble spots.

LTRS also know that infrastructure tends to draw out more cyclists. In this case this was true! However, there is apparently not accurate counts of the before and after cyclists which renders most of the research speculative. Further confounding evidence is the better reporting by the authorities of bicycle crashes. So there is some (good but confounding) reporting bias.

Not only that but 88% or a large majority of the crashes occurred while traffic is flowing aka stale green. BIKE BOXES AREN’T MEANT TO PROTECT YOU FROM FLOWING TRAFFIC. Thus, it’s a little inane to even focus on these crashes with relation to bike boxes at all.

The point of a bike box is for there to be a safe place to STOP during a red light and for you to be safe when starting up again. According to the linked research this goal was met.

Even more strange, the starting point for bike boxes places you WHERE VC CYCLISTS WOULD WANT YOU TO BE ANYWAY! Why VCers are so opposed to bike boxes is a mystery to me.

They make cyclists safer, they get more cyclists riding, and they reduce ambiguity. A win, win, win.

They do not help sell crappy classes and they are not part of Forester’s American Dream (sic).

Most motorists, post bike box, yielded to cyclists even when they were passing on the right (98%). This means that bike boxes are actually giving us the “Rights to the Road” that CABO and Co with their American Dream start are all too happy to piss away by their insane and meglanomaniacal obsession with eliminating cycling infrastructure.

Not I almost NEVER pass on the right unless there is a red light and I feel it’s safe to filter forward.

Since there are four problem intersections, the engineers are doing what good engineers do, they fix the problem by making their product better. They do not give up and teach us to get back into the cycling equivalent of caves.

Some of their ideas are to ban right turns all together on these streets. That should make accident prone behavior illegal. Other ideas include intersection reconfiguration by removing parking in some cases. Another idea is a separate signal phase for bicycles.

Despite this some idiot VCer misread this as to “prove” that bike boxes make things more dangerous:

” Yesterday, the city released a depressing letter (PDF) to the Federal Highway Administration that shows the bike boxes may have actually doubled the number of crashes.” (sic)

Actually this is not true. The bike boxes reduced the types of crashes that it intended to reduce and got more people cycling. The increased crashes could be from many reasons but there’s no clear evidence that bike boxes caused crashes: “coorelation does not equal causation.”

“What is the city going to do from here?”

If this genius actually read the report it’s spelled out. See above. Or better yet, read the full report yourself.

Some people see only what they want to see which is fascinating to me.

Reblogging: Sharrows Suck

June 3, 2013

Reblogging: Sharrows Suck

http://systemicfailure.wordpress.com/2012/11/14/sharrows-suck/

I have Erik Griswold to thank for this [https://twitter.com/erik_griswold]. Thanks, Erik.

According to the study: SHARROWS ARE MORE DANGEROUS TO CYCLISTS THAN NOTHING AT ALL. THAT IS, BY INSTALLING SHARROWS, CITIES INCREASE THE RISK TO CYCLISTS.

In fact, sharrows are riskier than sidewalk riding.

In the attached study, they compared doing nothing which they set as risk level 1 vs. other types of infastructre. Cycle tracks were rated at 0.05 or to put it another way, CYCLE TRACKS ARE TWENTY TIMES SAFER THAN DOING NOTHING AT ALL.

At 1.99, sharrows are twice as risky as doing nothing at all while sidewalk riding is 1.54.

Another intersting finding is that riding downhill is twice as dangerous as riding on flat ground. When riding down hill one is able to speed up without effort kind of like an e-scooter. Thus, I’d imagine that e-scooters are twice as dangerous as regular cycle. Further going up hill is twice as safe as riding on flat ground. Again, I believe that this is because the cyclists are forced to ride slower. This is part of my six word cycling safety speech:

“Ride slow, slow traffic, yield agressively.”

Further, the second line slow traffic is also proven by this paper as streets with 30 MPH traffic or less are twice as safe. Seems like a really dumb thing to put sharrows on high speed streets.

This has me reading further. I found a survey paper in which the authors read all the previous papers that they can find and make general conclusions about things:

http://www.biomedcentral.com/content/pdf/1476-069X-8-47.pdf

Here’s the conclusion:

“The principal trend that emerges from the papers reviewed
here is that clearly-marked, bike-specific facilities (i.e.
cycle tracks at roundabouts, bike routes, bike lanes, and
bike paths) were consistently shown to provide improved
safety for cyclists compared to on-road cycling with traffic
or off-road with pedestrians and other users.”

But whaaaaat about intersections? This is where accidents happen. Vehicular cycling should protect you, right?

No:

“It has been
suggested that the reason for high rates of bicycle-motor
vehicle collisions at intersections is that motor vehicle
drivers may be making “looked-but-failed-to-see” errors,
whereby they search for oncoming motor vehicles but do
not recognize that a cyclist is approaching because they
are not looking for them.”

Riding vehicularly is not going to get rid of the “looked-but-failed-to-see” errors.

Finally we get back to the tired and stupid helmet debate:

“The major advantage of infrastruc-
ture modifications, compared to helmet use, is that they
provide population-wide prevention of injury events
without requiring action by the users or repeated rein-
forcement. ”

Truth In Cycling

June 1, 2013

Over and over again, I hear snake oil claims in the cycling world.

I argue against them because I love to do so, but I figured that I should put my own standards for what is and is not true down in one place. This is not an exhaustive article, but it’s a starting point on what I think to be “true”.

First of all, I like studies. Just having you stuff in a pdf which is formatted like a scientific paper with charts and graphs makes me feel like I’m reading “data” instead of opinion. This is a bias of mine. I don’t think that biases are necessarily bad. I like to put my own biases out there; I like to be honest about these things. I expect the same level of honesty from others.

Back to studies: a good study is one in which it has a specific question which it asks. For example, for Vehicular Cycling: “Those who take cycling safety classes and practice its techniques consistently are less likely to sustain a serious injury or death in an collision with an automobile. This study will measure how much safer they are.” Then at the end of the study, we find out how much safer people are. We have a number that we can try to reproduce in other circumstances.

These studies are not hard to do. Relative to the amount of money we spend on classes and helmets and PSAs, these studies are highly inexpensive.

Another thing makes a study good is when one defines one’s terms ahead of time and gives good cut offs. For example, we hear that cycling is safe or unsafe or whatever. For me, the question is what does safe mean? Everyone uses this word, but few define it. I have looked for a definition and I accept WHO’s because it’s the only large organization which made such a study easy to find on the internet. Next time someone tells you something is safe or unsafe ask them how many people per 100,000 have to die or be seriously injured before you’d change your mind in what is or is not safe. If they don’t know then you are talking to someone who is merely speculating or they are blathering about their feelings. I like feelings and talking about them, but let’s not put our personal feelings out there like they are the truth.

Note, that we need to do all of this _before_ we gather data. There’s something about the human mind called confirmation bias which makes us look for data whihc confirms our position, only, once we are locked in. That’s why we need to get the standards created ahead of time. Otherwise, we’ll be tempted to move the goal post.

So when we safe helmets are “safer” what does that mean?

What’s a “helmet”? If we are going to compare helmets vs. non-helmets, we need to standardize our subjects on a particular make and model of helmet and talk about that. This is how real science works. In chemistry, we not only tell what chemicals we use, but we try to tell the manufacturer, too. This is because there are sometimes impurities in compounds which affect a reaction. Different impurities can have different effects. Different manufacturers can have different impurities. So if an experiment fails, it’s nice to look at who made our compounds and what the impurities are.

In pro-helmet studies, they never look at the differing manufacturers. This is because there are a bewildering array of different manufacturers with different helmets. This is like having a study which says that “vitamins are good for you” but we never look at which specific vitamins taken by which specific population and which specific doses. We just measure people who pop random pills vs. those who don’t. We don’t even give a defition of what “healthier” is. We do have a lot of graphs and numbers though based on circumstantial evidence. This study would be considered garbage in the medical world, but it’s the level of standard for our helmet studies where we measure children falling off their bicycles then we take that data and try to apply it to adults getting hit by high speed cars.

We’re not done! Next we need to look for confounding factors. For example, people who wear helmets are probably safer riders. We don’t know for sure, but they might be. One way to eliminate this confounding data is to pay people who normally would NOT wear helmets to wear them. Then we pay people who would wear them to not wear them.

Next we need to get a good statistician to make up the standards, as per above, on what is and is not acceptable. For example, if we see that there are 50% less visits to the ER by people who wear our standard helmet regardless of whether they were in our “safer riderss” category, we have a valuable claim. Otherwise, we don’t say that helmets are safer. I’m using 50% as an example. I don’t know how much safer helmets need to be because I’m not an expert in safety. I do know that no paper on helmets that I have read is up front on defining “safer standards.”

No safety equipment is safe 100% of the time. But we need to be safer than not in order to have a valid claim.

Finally, we conduct our experiment. We buy helmets from the same manufacturer and give them to cyclists to wear starting on the same date. Or we give cycling safety classes to those who normally would not take them. All of our cyclists wear wrist bands which tell the ER which data to collect and who to call with it.

THIS IS HOW A GOOD STUDY IS DONE! IF WE CAN’T AFFORD THIS THEN WE ARE SAYING SAFETY IS NOT NUMBER ONE. WE SPEND MILLIONS ON HELMETS AND CLASSES. WHY NOT SPEND A FEW THOUSAND TO ENSURE THAT THESE THINGS WORK AS ADVERTISED?

Until we see some good studies, using concepts as per above, let’s stop making outlandish safety claims. At best, we say that these things are unknown in their benefits. But for some reason when people see “unknown” they latch on to stupid nonsense like “better safe than sorry.”

No, we are better safe when we take an honest and sober approach to safety not when we waste money on questionable safety schemes. There are things which do make cyclists safer. They have been proven. Let’s spend all our money on those. Every last cent should go to things that we know that work.

Once we have reached diminishing returns which will be billions of dollars later, we can look into more marginal safety devices such as helmets and classes. But we need to be wise with our money. Otherwise, we’ll continue to get the same piss poor safety ratings we have had for the past few decades and our mode share will remain at a stangnant 1% which is where it should be based on the stupid and unproven notions which have been used to promote cycling safety.