Editor’s note

We love to give you the tools to keep you safe on your bike. New research from transportation planner and CyclingSavvy co-founder Mighk Wilson offers surprising insights about your safety when riding in bike lanes, on sidewalks, or on the edge of travel lanes.

In this three-part series, Mighk describes his bikeway research, and how the way he gathered data for it differs in critical ways from other bikeway studies.

You’ll be impressed.

Metro Orlando Bikeway Study

This is the first in a series of articles on new bikeway research which I’ve completed for my employer, MetroPlan Orlando. Findings of this research are useful for planners and designers, and for bicyclists. The study compared risks to bicyclists riding on sidewalks, on streets with bike lanes, and on streets without.

This first segment will cover:

• My own professional and personal history with bike lanes;
• How most bikeway studies don’t clearly show how bikeways might prevent motorist-caused crashes; and
• An overview of the data and basic findings of my research.

The second segment will help you make better decisions as a bicyclist. The third segment will address the “safety in numbers” premise and how bicyclists truly get to “Vision Zero” (the goal of eliminating fatal and serious injury crashes) [1].

Part One: Better Data, Better Understanding

I Used to Be Mr. Bike Lane

When I started work as a bicycle planner for MetroPlan Orlando in 1993, I was quite supportive of bike lanes. After all, the cities and towns that had lots of bicycle traffic had them. Wouldn’t that mean that people there had good experiences with them?

But it didn’t take me long to cross paths via internet forums with the infamous John Forester [2], and to have my assumptions challenged.

While Forester and his supporters had reasonable concerns about bike lanes, there was no solid data to show that they were worse (or better) for bicyclists than a regular travel lane. Concern about bike lanes was based mostly on direct experience. I had little, as the Orlando area had no bike lanes.

I found it frustrating that some bicyclists complained so much about bike lanes, while others expressed a strong preference for them.

A Quest for Answers

It took quite a few years for the Orlando area to get bike lanes and for me to gain enough experience with them. I found myself frequently having types of conflicts that rarely occurred when I used regular travel lanes. But still, I could find no good objective evidence about relative safety.

It became clear that to assess it, I would need lots of detailed crash data and good measures of bicyclist counts and behaviors.

Assembling crash data was fairly easy. MetroPlan Orlando had been collecting and analyzing crash reports since 1997. Also, I had a solid understanding of how bicycle/motor vehicle crashes happen.

But getting good counts would entail thousands of hours sitting next to roads, counting and observing bicyclists.

A few years ago, computer/video technology finally made counting efficient and effective. I now have the combination of crash, behavior and exposure data to allow useful analysis. But before I get to that, I want briefly to discuss the shortcomings of other studies.

Sloppy Bikeway Research

A number of studies published over the past decade or so have purported to show better safety performance of streets with bikeways compared to those without. Some are before-and-after studies of the same streets. These studies have a major problem: they treat all bicyclist-versus-motorist crashes the same, as if all would be affected in some way by the presence of a bikeway.

But clearly, the presence of a bikeway would not impact many crashes one way or the other. For example, a bicyclist might roll out of a driveway and fail to yield to an approaching car, or a motorist might blow through a red traffic signal at speed. Also, some crash types would likely be made more common by the presence of a bikeway, such as “wrong-way bicyclist” crashes when a bikeway either encourages or requires bicyclists to travel facing traffic.

Such studies have too many other kinds of failures to describe here, but the highly-touted 2011 Montreal cycle track study by Lusk and Furth [3] had a particularly serious failing. That study compared parallel streets with and without cycle tracks, but the paired streets were often radically different from one another.

One comparison, for example, was of a one-lane, one-way, low-volume residential street with a cycle track vs. a two-way, four lane street with high traffic volume and storefronts. A traffic engineering journal’s peer reviewers likely would have thrown out the Montreal study, but an injury prevention journal published it. The reviewers apparently didn’t catch this major failure of methodology.

Our Data

The MetroPlan study ensured that the control (no-bike-lane) streets were like the bike lane streets: same number of lanes, median type, same or similar posted speeds, similar traffic volumes, similar land use, and even similar surrounding populations.

We selected ten streets that had had bike lanes for at least ten years, and for each of them, a control street that also had not seen major changes for ten years. The streets were mostly suburban, with just a few a bit more urban.

We had ten years of crash data for all of the streets, categorized by crash type (who turned, who crossed, who violated right-of-way, etc.); the bicyclist’s position leading up to the crash (travel lane, bike lane, sidewalk or other non-roadway position); and the bicyclist’s direction of travel (with or facing the regular flow of vehicular traffic, or crossing the roadway).

Video Data Collection

Using a MioVision camera, we counted bicyclists, with their position and direction, on each bike-lane street and its control street during the same 48-hour period. With this information, we were able to estimate miles of bicyclist travel by multiplying the counts by the length of the study street, and then by 1,825, which gets us from 48 hours to ten years of exposure.

With ten years of crashes and of estimated exposure, we could calculate Bicyclist Miles Between Crashes. A high number means lower risk. (I’ve rounded to the nearest thousand miles for clarity.)

With the video, we were also able to estimate typical speeds for bicyclists. Traffic engineers typically use an “85th-percentile speed” for traffic studies. (85% of the travelers are going at or below this speed.) We found the 85th-percentile speed to be 12.4 MPH for sidewalk bicyclists, 15.7 MPH for bike-lane users, and 18.4 MPH for travel-lane users.

We also looked at five shared-use sidepaths (“trails” directly adjacent to roadways) that had been in place for more than ten years, and collected the same crash and exposure data for them.

Key Findings

This section will explore the risk of a motorist-caused crash.

Most Important Factor: Bicyclist Direction

It is legal to bike against traffic on a sidewalk or path, and illegal in a bike lane or travel lane, but regardless of bicyclist position, we found bicyclist direction to be the most important risk factor. The risk ratio was the same, 5.3 times greater, except for bike lanes, where it was 4.3 times. This means that bicyclists riding against the flow of motor vehicle traffic are 4.3 to 5.3 times more likely to be in a crash than those who ride with motor vehicle traffic flow.

These results show higher relative risk than prior studies, but this study had many more streets and better bicyclist-count data. (Wachtel and Lewiston in 1993 [4] found 3.6 times greater risk, and Huang and Petritsch in 2007 [5] found 4.4 times greater risk.)

Risk by Position

So we know that going with the flow is much, much safer. But if we’re going with the flow, is it better to be along the edge of a travel lane, in a bike lane, or on a sidewalk? For bicyclists traveling with the flow, the Miles Between Motorist-Caused Crashes were:

Travel Lane Edge – 31,000 Miles (Highest Risk)

Bike Lane – 64,000 Miles

Sidewalk – 122,000 Miles (Lowest Risk)

I have assumed that the bicyclist in the travel lane is riding along its right edge. Very few bicyclists use lane control, and only a tiny percentage of crashes involves bicyclists using it. This study could not assess that strategy.

These numbers make it look like the sidewalk is the safest place to ride, with four times lower risk than the edge of the travel lane. The sidewalk looks better than the bike lane too. But stay with me…

The four main motorist-caused crash types for bicyclists going with the flow were: overtaking motorist, drive-out, right hook, and left cross. (Dooring was not a significant issue with these streets; only two of the twenty streets had parallel on-street parking. There was only one reported dooring during the ten-year period.)

Overtaking crashes are very rare. Of 428 motorist-caused crashes on these twenty streets, only ten (2%) involved overtaking motorists. Six of those ten involved bicyclists in bike lanes. But the bike lanes did show a much lower risk for overtaking crashes: 585,000 miles between crashes compared to 92,000 miles for travel-lane bicyclists.

Setting Overtaking Crashes Aside

For now, let’s set overtaking crashes aside and look at the risks for drive-outs, right hooks and left crosses. Again, we consider only bicyclists going with the flow.

Travel Lane Edge – 61,000 Miles (Highest Risk)

Bike Lane – 75,000 Miles

Sidewalk – 122,000 Miles (Lowest Risk)

Here we still see much lower risks for sidewalks, and somewhat lower for bike lanes. But why would there be lower risks for those crash types? They all occur at intersections and driveways with no sort of “protection” for the bicyclist. Don’t experienced bicyclists avoid using sidewalks — and sometimes even bike lanes — precisely to avoid such conflicts?

I have the answer for you in the next article.

---

Footnotes

[1] Wikipedia article describing Vision Zero, with links to other resources.

[2] John Forester, pioneering bicycling educator, died in April 2020. We have an evenhanded article about him.

[3] The Montreal study may be found here. Michael Kary’s critique of it may be found here, Wayne Pein’s, here and Paul Schimek’s here.

[4] The Wachtel and Lewiston study is available online.

[5] The Huang and Petritsch study is available on the Metroplan Orlando site.