I come in peace, with a bike helmet on

ATTENTION CONSERVATION NOTICE: I come in peace with my bike helmet. Liking bike helmets does not mean I think they substitute for street design or culture, and I really doubt anybody else who isn’t crazy thinks helmets are a substitute for design or safety in numbers, either.

Ok, so brilliant student Shane Phillips (who blogs brilliantly here) yesterday shared something about bike helmets and design, and the piece suggested that bike prevalence is a safety substitute for helmets, and I said, hey, it would work for me if we had both bike prevalence and helmets. When it comes to most things regarding safety, I am more Adrian Monk than Laura Croft.

Note the metaphors there: Adrian Monk is not particularly rational about his assessment of risk. I’m not either. I’ve known way too many people I’ve loved who have been hurt by bike crashes, and one of the worst did not involve a vehicle, at all. I’m paranoid. Every day is a long list of worst-case scenarios that float through my head, and bike injuries are in there with falling meteors and Donald Trump becoming president.

You do realize I think the NFL should change its rules to touch football, right?

So anyway, what I thought was some good-hearted banter turned into a boy urbanist coming in mansplain ABOUT THE FACTS, AND THE FACTS ARE THAT BIKING IS MUCH SAFER THAN BEING A CAR AND SO I GUESS HE TOLD ME.

But I never said that driving was safer. Nobody had. It was an attitude he assumed I had because he wanted me to be wrong, and he wanted to yell and scream and Be the Smartest Boy Urbanist in the Room.

Then another guy piped in, and he at least was polite. But his statements made me uncomfortable, too, because he trotted out the whole “Denmark totally proves that helmets have zero correlation with bike fatalities” conclusion “because almost nobody wears a helmet but there are few fatalities.”

Now, I have nothing against the Netherlands or Denmark, but we can’t use these numbers in this way. I’ve heard my students quote this idea, and while it’s not the worst instance of poor urban numbers, it’s not rigorous thinking about how things might relate.

The idea seems to go something like this:

1) People don’t like bike helmets and that’s what keeps them from riding a bike;

2) Bicyclists are safer when there are more bicyclists are on the road (“the safety in numbers” effect); and thus

3) If helmets keep bicyclists off the road, they do not improve safety because they dampen the “safety in numbers” effect.

Now, most people can understand the reverse causation possibilities in 2: bicyclists are safer where there are more bikes and it’s likely there are more bikes where things are safer, either because of design or culture or both, which is what we likely see in Amsterdam.

And then the corollary: the more bikes there are, the fewer cars there likely are, and the lower fatalities are more likely to occur because of the latter, not the former. The former is a lever for the latter, but they are still two effects.

That part, I totally buy. It’s the helmet part I’m having trouble with. Why? Because for years I’ve advocated for and studied transit, and I’ve heard people tell me every specious excuse in the book for why they don’t take transit. Them: “It doesn’t come near my house.” Me: There’s a bus stop on the corner, a five minute walk.” Them: “Oh, but that’s a bus. Buses take too long.” Me: “It’s a 15 minute trip, you have no transfers.” Them: “It doesn’t leave when I want it to/the sun gets in my eyes/I’m allergic to the bus seats/people look at me funny/I got sick on a bus once after drinking seven bots of tequila so I just know I get bus sick.”

So if you aren’t riding a bike because you don’t like wearing a helmet, then you are a fair weather bicyclist in my book. “It messes up my hair.” Buy a comb. And what? You don’t ride if it’s windy? When I actually get my fat self on a bike, which isn’t often, and I’m not wearing a helmet, I always wind up looking like I did my hair with an immersion blender anyway. Hair, schmair.

Now, in fairness, I do understand the problems that people have with helmet laws. So you forgot your helmet, and then what? You get a citation? That’s a hassle, just like having to have exact change for the bus is a hassle. But how strictly are helmet laws really enforced? Anybody got an idea on how strict enforcement is in places where they have helmet laws?

BTW, I don’t think helmets should be mandated because it just strikes me as overreach. That’s different from wishing people would wear helmets.

More to the point about research, I can find no believable research that demonstrates “safety in numbers” results from not encouraging helmets, and I also can not find any credible data that “safety in numbers” precludes wearing helmets or that helmet laws discourage street design, or any other such thing.

Here’s the story that circulated on Streetsblog, where the columnist, Angie Schmitt, who is really a wonderful columnist I often read for Streetsblog, concludes things I really don’t think you can with the numbers from the graphic she presents.

Let’s start with the title. I know titles are always screwed up, but the title starts us off in ways that make me squirm. Helmets are not the answer the bike safety. Well, did anybody say they were “the” answer? Can somebody point me to the article or expert saying “We don’t need bike lanes because we have bike helmets?” Is somebody saying that? Did Congress slash funding for bike facilities because we have helmets now? Because that would be crazy if people believed those things, and I’m willing to believe that some people say this because some people also think that aliens embedded a code in the Torah. But…really?

Then there’s the next line:

Better street design and getting more people on bikes — not blind faith in helmets — are the keys to making cycling safer, recent research has shown.

Are there people out there in the world telling people that all they have to do is put on a bike helmet and they should be fine, just fine, even if they peddle themselves onto the 405 when traffic is moving at 60 mph? Because that would be nut bar. Who is this person, or these people? Or this just shadowboxing to set up a column?

Just about everybody who does safety research groks that safety isn’t a one-factor deal, except when it comes to very simplistic things, and mobility systems are not simple. Seat belts are fine, but seat belts, crumple fenders, and airbags are a whole package. Seat belts, crumple zones, airbags–and speed limits and proper sight distances and drunk driving education are still yet more potential dimensions and variables.

You know what else increases vehicle safety, a lot? Using cars less overall. I’m a fan of that one, too. But that doesn’t mean I think we should get rid of airbags.

Granted this, why do some US bike advocates seem to think that if you favor wearing helmets, you are against good bike design? Why can’t people have both?

The logic strikes me as a bit like this: You think people should eat right. YOU MUST BE AGAINST EXERCISE THEN. WHY DON’T YOU PEOPLE WHO WANT TO EAT RIGHT UNDERSTAND THE IMPORTANCE OF EXERCISING JEEZ? Shit, some people hate to exercise, and if you associate exercise with eating right, people won’t eat right because that means they’ll have to exercise.

Um, no, it’s possible that both are a good idea, and that both contribute, even if one effect is more important than the other.

Just like it’s possible that there is a “safety in numbers” effect and a helmet effect on safety.

Then we move into the rest of the column, which shows this graphic, which is supposed to show us that bike helmets do not contribute to safety because look at

Cj L75WVEAADWl 1

This graphic, like the one one she references but does not display, I think came from Toole Design Group.

Now, Schmitt grants that these are only eight data points, but then goes on to say:

Of these countries, the U.S. has the highest rate of helmet usage among cyclists — around 55 percent — but also the highest cyclist fatality rate per distance traveled. Meanwhile, in the Netherlands, where helmet use is practically nil, cycling is much, much safer.

While this is just eight data points, higher helmet use seems to be associated with higher fatality rates. Intuitively, that makes some sense. The more dangerous an activity, the more people feel inclined to take steps to protect themselves.

So the url says this is “why helmets are not the answer in one chart”…but this had better not be the basis for the conclusions because this chart does not say anything consistent about anything. Technically the charts shows sixteen data points, for the anal data types, and 8 cases.

One point of contention: helmet use is not practically nil in Denmark–her idea of practically nil and mine are not the same; and with helmet use, I suspect that there are strong nonlinearities by age, cycling experience, and gender. You don’t necessarily need everybody wearing a helmet. It may be that the Netherlands has the right people wearing them, while all the low-probability injuries and fatalities types go without. Sure, it means not everybody wears a helmet, but it also could mean that in individual incidents, helmets help.

But I digress. The major point: why are we highlighting Denmark/Netherlands and the United States here? Because if I want to make the opposite claim–that bike helmets clearly do help with safety–then I have those data in the chart, too: compare Finland and the UK. Or Finland and France.

If you wanted me to do actually my job here, as a researcher (ugggghhhhhhh work), I’d say we cannot use these data, at all. They are noninformative about correlations because aggregating bike safety at the national level isn’t really useful. Nation-states do not represent biking environments or cultures, per se.

We all know that the US has poor infrastructure and design for bicyclists in many US cities, and since Denmark and the Netherlands are highly urbanized, you can maaaaaybe use the nations as proxies for bicycling environments there. Maybe. But as some cities in the US improve their design, the US as whole becomes a very poor proxy. Even as it is, freight volumes and practices differ, weather, etc etc between Denmark and the US, and it’s hard to compare rates across a big country and small country because small countries are always going to have more variation in aggregate numbers than the big ones. I don’t think large numbers problems explain the differences between the US and the Netherlands and Denmark–I think the latter’s planning clearly demonstrates effectiveness–but it does make measuring the effect more of challenge.

We also have the problem that fatalities might be even lower than they are–closer to my practically nil than Schmitt’s practically nil–if the Dutch wore more helmets more frequently. The comparison that matters is between helmet wearers versus non wearers in Amsterdam. As one commenter noted: “Show me a statistic of fatalities of helmet wearers who crash on protected multi-use bike paths.”

THAT is the world I want to live in: helmet wearers wrapped in bubble wrap riding great bikes on great bike facilities. Maybe we could even have some these cool bubble jobbies to wear when biking:

Bubble soccer USA Rent Bubble Soccer
(Image from http://www.bubblesoccerusa.com; God that looks fun.)

What matters are the practices and designs in place. In order to know what we’d like to know about the independent effects of design and helmets, we need to know about specific intersections, specific street set ups, specific traffic conditions, and specific crashes.

And that’s a huge problem. Because–and this is a good thing–bike fatalities are statistically rare events, even in the US, which means doing the sort of research you’d like to do–controlling for all the design and personal safety factors (like wearing a helmet) that enter into fatalities–requires data usually too sparse to grant your statistical analysis the power you need to find effects.

The other problem with these conclusions is that helmets may not be useful in preventing fatalities, but they may be useful in preventing injuries. Injury prevention has social value, too; and it may be that “safety in numbers” would also influence injuries for the better concurrently with fatalities, or it may be the opposite, just like on some freeway lengths, lower speeds are associated with more crashes but fewer fatalities. (Congestion lowers speeds, therefore decreasing severity but increasing frequency.) I am betting bike-related injuries follow similar nonlinearities into zones where you might have more frequent incidents (bike-ped, bike-bike), but that those incidents are not as severe as what happens when there are too damn many cars moving too fast.

The other study that Schmitt highlights is better, but still has some pretty troubling data problems Here is a pdf to the study. This is a very nice study, and the authors are explicit: they state they inferring and do not have causality, and–and I think this is interesting–gender winds up being a significant variable. I am not sure I understand their interpretation, but maybe I will see it more clearly when I reread it.

The problem here–and I’m not saying anything the authors do not know-is that they use bike helmet laws as a proxy for wearing bike helmets, and that is not optimal. They may have no other choice, but self-reports are a problem. It is a truth universally acknowledged that human beings are bad self-reporters, on everything, but particularly when they know there is a “right” and “wrong” answer, like safety and health questions.

They are two steps removed from being able to make policy conclusions about helmets. First, do helmet laws increase helmet use, and then bicyclist who experience trauma wear helmets? And another problem: if bike helmet laws do suppress use, it would exert an effect both on overall amounts of cycling which is likely to (though not inevitably) lower incidence and on the safety in numbers effect. The endogeneity problems make things tough, right along with the rarity of the event that prompts people to aggregate the data to non informative spatial units.

Again, we go back to the problem of really wanting to be able to look at specific traits–of places, people, behavior–of specific crashes and compare them with nonfatal incidents in those locations as well as overall exposures at those places to avoid the Abraham Wald survivorship problem .

Keep in mind that I am not saying that we do not need better bike design. Bikes and bike facilities more than warrant priority. People enjoy them, they promote health, and they do not cost very much granted what they can deliver in terms of mobility and enjoyment. Promoting cycling and walking are no-brainers. Do I think they solve climate change? Eh, maybe. Not convinced. But if people enjoy these activities, then why wouldn’t we make space for them cities based on that rationale alone?

Schmitt’s last comment is what inspired me to write the blog post:

Despite the high rate of helmet use in the U.S., helmet campaigns have clearly failed to make cycling as safe as it should be. If anything, they’ve distracted from the much more important work of designing safer streets and reducing motor vehicle speeds in cities.

Now, wait. Suggesting that you slap a helmet on your head, which takes all of 30 seconds, distracts the US from doing a good job with bikes? Come on.

Really, folks, people like me are not your enemy. I just want you to keep the skin on your face where it is if you get unlucky one day. If you don’t want to wear one, well, that’s up you. But just like the people who love me wish I’d lose weight, I want you to ride safe, too. I want you to do what you can to protect yourself as US cities try to get their shit together and supply better spaces for riders.