No Surprise: Study Finds Sharrows Don’t Make Streets Safer

The study basically confirms what any cyclist who has ridden in sharrows already knows.

Written by:

Sharrows: clip art-style bicycles with arrows painted onto the roads to indicate where bicyclists should ride on a street that remains, ultimately, dedicated to automobile traffic. Sharrows are what cities install when they want to appear as though they care about bicycling, but can’t or don’t want to muster the political will to actually change anything significant in its favor.

It has long been assumed by bike advocates and everyday riders that sharrows do very little, if anything, to increase road safety for people on bikes. As it turns out, those assumptions were correct.

A recent study undertaken by University of Colorado Denver researchers Nick Ferenchak and Wesley Marshall examined safety outcomes for areas of Chicago that received bike lanes, sharrows, or no bicycling infrastructure at all. The study was conducted before Chicago had much in the way of protected bike lanes, so in this case no distinction was made between types of bike lanes. The study concluded that, while bike lanes encourage more people to ride and lead to increased safety for people on bikes, sharrows do neither.

The study was conducted by dividing Chicago into three categories: areas where bike lanes were added, areas where sharrows were added, and areas where no bike infrastructure was added, each between 2008 and 2010. Ferenchak and Marshall then analyzed rates of biking and rates of cycling injuries in each area to determine the effects of the road treatments or lack thereof.

They found that rates of cycling more than doubled in areas where new bike lanes were added, compared to only a 27 percent increase in areas where sharrows were added. Interestingly, there was a 43 percent increase in areas where nothing changed. In areas where new bike lanes were added, the rates of injuries to people on bikes decreased 42 percent, compared to only a 20 percent reduction in areas with sharrows, which was worse than the 36 percent reduction in areas where nothing was changed.

While this is only theorizing, part of the reason injuries decreased more in areas where nothing happened than in areas with sharrows could be due to sharrows’ ability to change cyclist’s behavior without actually changing circumstance. For instance, sharrows may convince bicyclists to ride on streets they wouldn’t otherwise ride on because of an increased perception of safety, while not actually providing any real protection or providing any impetus for drivers to change their behavior accordingly.

Dutch bike planner Dick Van Veen once told Streetsblog that putting sharrows down on a fast street with no corresponding traffic-calming measures would be “unethical.” If you’ve ever ridden down a fast-moving, busy street in the sharrow line, you realize quite quickly that most drivers are not treating those painted arrows as proper infrastructure.

This most recent study basically confirms what is quite easy to feel when you’re out in the sharrows on your bike. Bike infrastructure needs to be safer than this, it isn’t enough to simply paint an arrow on the ground and call it a day. “I think our main takeaway is that we need appropriate infrastructure,” Ferenchak said. “Sharrows don’t dedicate any space to bicyclists.”



  • Dave Thielk

    I think the value of sharrows depends a lot on many other factors. My experience suggests that if cars are slowed to 20mph, and there are significant traffic calming measures, then sharrows are a reminder that bicyclists are in the roadway also. But when speed limits are 30mph and above, drivers just muscle their way past, often angry at the cyclists. We have sharrows put in on a heavy use roadway (single lane) by our city engineer. The speed limit is 40, and then down to 30. The drivers just roar past at 45 to 50, often appearing to drive cyclists on to the shoulder. The drivers are engraged by the sharrows. And because the speed limit is high, drivers do not feel obligated to slow down. And further, they are resentful that they should even have to wait for the opposite lane to clear so that they can pass safely.

    Sharrows are best as simple reminders when all of the other infrastructure is already in place.

  • Patricia Kovacs’s comprehensive review of the Frenechak-Marshall study is here:

  • Augsburg

    Although I can understand the consternation behind this article – drawing the conclusion that the study proves what “all cyclists” already know. That shadows are a feel good for the politicians, and do no real good for cyclists. As a long time cyclists putting about 2,500 miles on the road each year, I have to disagree.

    First of all, no study of roadway safety in the US, prepared using collision or “accident” data is worth much these days. Most major cities gave up reporting collisions unless serious injury or death occurs. The vast majority of conflicts and less serious collisions are never reported or captured in the data.

    Secondly, As someone that puts many miles on roads with and without shadows, I my conclusion is they do have some effect. Not a dramatic difference, but a difference. For example, I will experience less aggressive behavior fro drivers of motorized vehicles on roads with shadows. Less honks, fewer shouts out the window, less motor revving and zooming by too close. The shadows do remind drivers to share the road. Without them, drivers are often complete a$$holes.

    • Mirsa

      The article makes no mention of how the cyclists are using these sharrows. If a cyclist hugs the curb, it is essentially no more useful than a narrow bike lane next to a substandard width lane, and drivers will inevitably attempt to drive within the same lane with you.

  • I agree that the study is suspect. In Long Beach we have put sharrows on several streets. Of particular note are the green sharrows (which the FHWA unfortunately no longer authorizes) on 2nd Street in the commercial district of Belmont Shore and on4th street in the RetroRow commercial area. Our studies have shown both of these to be very successful. On 2nd Street we saw a greater than 100% increase in the number of cyclist using the street, a 75% decrease in the number of cyclists riding on the sidewalk (which was a major complaint of the business owners), a marked decrease in the number of cyclists riding in the door zone, the number of crashes stayed the same (even with the doubling of the number of riders) and significantly a lot more understanding by motorists of where the cyclists are allowed to (and suppose to) ride. On 4th Street we used closely spaced standard sharrows and have seen similar results. Both of these streets have 25mph speed limits and are heavily travelled by both cars and bikes. What I think is most notable on both streets is (1) the increase in the number of riders, (2) the decrease in the number of cyclists riding on the sidewalk, (3) the decrease in the number of cyclists in the door zone, where they are not only in danger of being doored, but are also much less visible, (4) drivers are much more courteous and finally (5) no increase in crashes even with the significant increase in the number of riders.

    Please…come to Long Beach and test out our well designed sharrows along with our bike boulevards (which also use sharrows), our separated lanes and our class 1 beach path where we separate the bike and pedestrian traffic.

  • Those are like “Share the Road” signs, only they should add “when I’m not on it”.

  • Frank Krygowski

    I’ve read the study, although it was a difficult exercise. Why? Because the “nonsense” frequency gave me a headache from all the eye-rolling I did!

    Example: The authors begin by claiming the only initial motivation for sharrows was to reduce dooring. That’s absolutely false. They then claim that dooring is a rare problem, despite saying that in Boston, doorings were 11% of car-bike crashes, and 29% of bike messenger-car crashes. Other sources say 20% of Chicago bike crashes are doorings. Those numbers don’t qualify as “relatively rare and benign.” And the author’s polemic question, “Why would we install sharrows that avoid doorings at the cost of of increasing risk for vehicle collisions?” is nonsense, because it displays the author’s hidden assumption that riding more prominently in the lane increases risk of collisions. Most truly competent cyclists have found the opposite to be true, and _that_ is a major point of sharrows!

    Another example: The authors then suddenly switch to claiming that the intent of sharrows is (instead?) to increase passing clearance by cars. They claim the results are “highly variable,” but cite several studies in which that spacing was, indeed, found to increase, by anything from several inches up to 2 feet. They also mention studies correlating sharrows with less sidewalk riding, greatest benefit on four-lane roads, etc. Yet they disparage sharrows as ineffective.

    It’s significant that the paper does not ever count actual cyclists using actual sharrows! (Nor bike lanes, nor plain streets, for that matter.) Instead, it uses approximate data on bike _commuting_ as a proxy for total ridership. And it generates its risk ratios by using _all_ cyclist injuries. In other words, a wrong-way midnight drunk’s crash is divided by the number who claim in surveys to be commuters.

    And not the number of commuters on the street where the drunk crashed! The authors use approximate data on city “block areas” of unspecified size. If a “block area” has some streets with nothing special and one street with a short section of sharrows, it apparently counts as a sharrow “block area.” Furthermore, the before-after computations that purported to evaluate the cycling treatments don’t use even the same “block areas.” Those boundaries shifted over the study period, so the areas are only approximately similar.

    Given all those shortcomings: What the authors actually found was that “block areas” with sharrows saw an increase in bike mode share (based on surveys, not on observation) from 0.69% to 1.32%. As with “block areas” with no changes (0.33% to 0.75%) or those with bike lanes OR recreational bike paths (0.64% to 1.78%), those changes might best be characterized as “from negligible to negligible.” However, all are increases.

    Likewise, the injuries (of _all_ cyclists) per 100 bicycle commuters dropped from 33.3 to 21.1 for no treatment, from 31.2 to 25.1 for sharrow-containing “block areas,” and from 59.2 to 34.4 for bike lane “block areas.”

    Now think about those magnitudes. 59.2 reported injuries for every 100 bike commuters! How can that be? Again, its’ _all_ reported bike injuries – the drunks, the kids falling off their bikes and rushed to ER by helicopter parents, the wrong-way-sidewalk riders, the stunt riders and more. It makes sense only because the commuters are a negligible portion of the total – yet that negligible portion is used as a denominator in evaluating safety. The logic is mathematically weird.

    Note that the sharrow “block areas” did experience an increase in cycle commuting, as well as a decrease in (computed) crash rates. How, then, do the authors justify their statement that “sharrows have less than desirable outcomes”?

    The authors do a fair job of listing the shortcomings of their methodology: the assumption that the impact of bike infrastructure would be experienced _thoughout_ a “block area” if any street gets some bike facility; the assumption that bike commuter survey responses accurately indicate _total_ bike use. They admit that the proper way to evaluate both ridership effects and safety effects would be by accurate before-after counts, but say they chose their strange methods because – well, because it was easier to get this data!

    But they also hint at perhaps the biggest shortcoming: the implied direction of causality. The entire paper implies that the bike lanes _caused_ more increase in cycling than the sharrows, and _caused_ greater increases in safety. Segregation-promoting websites like Streetsblog have certainly taken up that cry. However, four paragraphs before the end, they say “the results of this research do not imply causality…”

    Right. In effect, its an admission that their paper’s message is not justified by their own data, as weak as that data is.

    To explain: There could be many undetected or ignored confounding factors behind the data. Did bike lanes increase cycling more where they were installed? Or were the bike lanes installed because those “block areas” had lots of cycling, good cycling conditions and lots of commuting potential via short trip lengths, hip local culture, accommodating businesses, etc.? Did those already cycling, and right on the edge of commuting, demand the bike lanes? Would the commuting have increased just as much during this period of growing bike fashion? We don’t know; but it would not be at all surprising if DOTs installed bike lanes where cycling was growing the fastest and requested the most – in other words, where cycling was already pretty good.

    Finally, the authors tip their hand in their final three paragraphs. These young research assistants in Denver constructed this paper because, according to them, Denver is on track to completing its planned sharrow installations, but it is “falling behind” on other facilities. (Yes, gentlemen, installing any bike lane, especially a “protected” one, is more time consuming because of the need for acquiring and clearing right-of-way, or studying the effects of reducing traffic lanes. Separate bike trails are even more complicated. Who could be surprised that sharrow installation happens more quickly?)

    Here’s their final polemic: “It is time that sharrows are exposed for what they really are, a cheap alternative that not only fails to solve a pressing safety issue, but actually makes the issue worse through a sense of false security.”

    That’s what I call a “Danger! Danger!” argument. “Oooh, riding a bike is SO risky that it’s a pressing safety problem!”

    Let’s keep in mind that data shows Americans ride over 10 million miles per year between fatalities. And that, per mile traveled, those walking experience three times as many fatalities as those riding bikes. Those numbers, BTW, come from John Pucher, who – despite the data’s message – is another “Danger! Danger!” proclaimer regarding bicycling.

    It is flatly anti-cycling to exaggerate the danger of bicycling, and to claim that one dare not cycle until one has completely segregated infrastructure. It suppresses cycling here and now, and tends to blame those few victims who are injured while riding legally, using their rights to the road. “Advocates” of that stripe need to be exposed for the fear mongers they are.

    Meanwhile, the main benefits of this paper are exposing those who buy its message. They place themselves firmly in the fear-mongering camp, and illustrate their gullibility and their bias.

  • Naked point estimates are completely useless.

    We have no idea how much statistical confidence we have in these estimates. For example, a confidence interval of +/- 10% is not uncommon for small studies, if true there would be no statistical significant difference between sharrows and no change in terms of injury rates. And all the pontificating about cyclist behavior? It would be pointless speculation based on no real data (i.e., “truthiness”)… which seems to be the approach of most “journalists” these days.

    Lazy journalism boys and girls! Come on step up your game!

  • Patricia Kovacs

    What happened to the number of dooring crashes that are being tracked in Chicago? Through 2012, there were an extraordinary number of doorings, and a recent article from 2014 says 1 in 4 crashes are doorings. These are NOT tracked by Illinois DOT, which is where the safety data was retrieved.
    I would not call these crashes “benign”.

  • Can you provide a link to the study please?

  • Patricia Kovacs

    The study is now available from the TRB for $20. I just read it. They formed their conclusions about the impact on cycling and crashes based on the commuting data from census data and crash reports from the City of Chicago. But the cyclists may not have been riding on the roads with the infrastructure and they may not have been injured on the roads with the infrastructure. Shouldn’t a study look at before and after bike counts and crash data on the streets being studied? The census groups with the bike lanes also have sharrows. All the census groups have street without any infrastructure. Maybe those were the safest place to ride, who knows? I would like Momentum Mag to read this study and then comment on it.

    • jack hughes

      MOMENTUM is rather unlikely to give the study an unbiased review. The ‘scare’ conclusions about shared lane markings is too much in keeping with the magazine’s prominent bias toward infrastructure “solutions” at all costs.

  • Patricia Kovacs

    This surprises me since all previous studies show an increase in cycling when sharrows are added. I’m wondering if you read the study? I can’t find it anywhere. As I asked on another blog where this was posted, is the DOJ going to look into this practice of putting bike lanes in one census area and nothing in another? Here’s some data from Columbus, Ohio. Sharrows were added to a 6.5 mile section of High St in 2010-11. Crashes in that section of road stayed the same before/after sharrows. But MPO bike count studies showed an increase from an average of 21 to an average of 39 cyclists in the 2 hour collection periods. Another location showed an increase from an average of 9 to an average of 19. That’s about double, right? Sharrows do change behavior, if done correctly. They show cyclists where to ride, away from door zones, in the center of lanes, in the proper direction. They also show motorists that bikes belong. So let’s see the study. Let’s see how the authors came to their conclusions.

Autumn Gear Guide

Find inspiration in our Gear Guide that will keep you out on your bike through wind or rain.

Download Now