Interpreting Interpretations: An Analysis of the Bike Accident Tracker

About two weeks ago, Zusha Elinson of The Bay Citizen published an article detailing the "Bike Accident Tracker" that his team developed to map and analyze accidents involving cyclists in San Francisco.   There are a number of really incredible aspects of this data app as well as a few questions yet to be answered regarding its utility.

First, the good stuff.   Harmonizing accident data is an important step in drawing attention to potential problem areas within a city.   The data was supplied by the San Francisco Police Department, and includes collisions reported between January 1, 2009 and November 30, 2010.   Even if some accidents go unreported, the 1,147 accidents that were reported for this period, when mapped, begin to depict patterns.   For example, the Mission has the highest number of accidents per year, more than twice the number of collisions than the Financial District.

People also have an opportunity to add unreported accidents to the map, but smartly, these entries are all reviewed before being posted on the map, and user-submitted data is clearly distinguished from SFPD data.

Bike Accident Tracker
Image Credit: The Bay Citizen

The report also goes a bit deeper than simply mapping the data.   The leading causes of accidents involving cyclists are included in the analysis, as is information regarding who was at fault (cyclist vs. motorist).   Cyclists are reported as being at fault 50% of the time, most often for speeding (yes, cyclists speed, too) and disobeying red lights and stop signs.   Motorists are also guilty of speeding, failing to signal, and dooring.

This is all great information to track and scrutinize.   But, without ridership information to analyze in conjunction with the accident data, how useful is the Bicycle Accident Tracker?   The San Francisco Municipal Transportation Agency does collect statistics on ridership, but its counts are based on a single day of manual counting in August.   While there are some useful trends noted in the SFMTA annual report, such as identifying a 58% increase in observed cyclists since 2006, it is difficult to draw conclusions from one day of counting each year.


In Arlington, Virginia, bike counts are performed over a three-day period every quarter.   In other cities, including Portland, automated counters have been installed to measure traffic.   Of course, more frequent counts and automated systems are more expensive than annual counts, but the National Bicycle and Pedestrian Documentation Project correctly states, "Without accurate and consistent demand and usable figures, it is difficult to measure the positive benefits of investments in these modes, especially when compared to the other transportation modes such as the private automobile."   Similarly, it is difficult to properly assess problem areas without accurate ridership data.

According to Elinson, "our hope is that the tracker is used on many different levels: by policy makers and transportation planners when making decisions about street design and transportation policies and by cyclists and drivers to understand dangerous spots in the city and how to make them safer."   I hope so, too.   I also hope that, whether it is Elinson's team, the SFMTA or a third party, more comprehensive data about ridership can be collected and reviewed in conjunction with the accident tracker.


The Mission has more than double the number of annual accidents as the Financial District, but is the Mission District a more or less safe place to ride a bike?   If the Mission has ten times the number of cyclists annually but only twice as many accidents involving cyclists, then what do the accident statistics really indicate?

The Bike Accident Tracker is an important tool.   The tracker can “help move everyone past the overly-simplified arguments about cars vs. bikes and spur discussions on bike safety that are based on data instead of preconceived biases,” says Elinson.   And he’s right–let’s use the data, including the information about what causes accidents, to be more conscientious cyclists.

It is vital that we make the best use of the data that we have regarding accident locations and causes of accidents, but it is equally important to explore ways to improve the usefulness of the tracker by collecting and overlaying ridership statistics.   When advocates take this tool to the powers that be in San Francisco or in other cities, then they can not only point out problem areas, but also they can demonstrate demand, context and criticality.

Post navigation