Measuring Detroit’s “Bike Friendliness”

  • The current “bike friendly” city measurements and rankings are largely based on heavily flawed and inaccurate bike commuting data.
  • Bike commuting data does not represent actual biking levels in cities like Detroit where a majority of workers travel to the suburbs for their jobs.
  • Relying on bike commuting data ignores the majority of other bicycle trips made within cities.
  • Other Detroit data sources can be a more accurate measure of bicycle friendliness.

We’ve been working with the Detroit Office of Sustainability on how bicycling, walking, micro-mobility, bike lanes and greenways fit within their planning efforts. They want to measure Detroit’s progress in these areas. Initially they’d suggesting using ratings from national bicycling organizations, but those are highly inaccurate and rate Detroit poorly. Those ratings clearly do not reflect the reality of Detroit’s diverse bicycle culture that includes the largest weekly bike ride, the most bike clubs, and the second largest protected bike lane network in the U.S. This article explains why these ratings don’t work and provides better data options for measuring progress.

Measuring Detroit’s “Bike Friendliness”

There is no U.S. standard for measuring the bike friendliness of the city. One could expect the number of people bicycling in a city to be a good reflection of its bike friendliness. However, such data does not exist.

Despite this, many national organizations receive grant and private funding to rate U.S. city bicycle friendliness. They rely on heavily flawed data that impacts Detroit’s measure to a much greater extent than other cities.

American Community Survey (ACS)

National groups largely base their bike friendliness rating on the number of workers whose primary means of transportation to work is bicycling, which is from the Census Bureau’s ACS. This is a highly flawed approach since:

  • Commuting to work makes up minority of all trips (~16% nationally)
  • ACS only surveys U.S. citizens with a residential mailing address
  • ACS under counts bicycling trips, e.g. multi-mode trips, non-primary transportation modes, seasonal transportation modes, etc.
  • ACS only counts trips for those that have jobs
  • ACS data has high-error rates due to the low sampling rate within cities like Detroit

The other large issue is a majority of Detroiters have jobs outside of the city, making bicycling to work an unrealistic option. [UPDATE: Detroit also has a below average response rate to Census Bureau surveys and 86% of the population live in “hard-to-count neighborhoods” per Associated Press analysis.]

Even though the data quality is a poor measure of commuting modes, it’s available nationally and easy to extrapolate. For example, in 2016 the League of American Bicyclists ranked cities by bicyclist fatalities per 10,000 bicycling commuters. Due to our low commute rate, Detroit was rated the second most dangerous U.S. city despite none of the fatalities being bicyclists commuters.

We have made groups like the League of American Bicyclists and People for Bikes aware of this issue, but we have not seen any improvement in their methodology. For this reason, we don’t assist these groups when compiling data from Detroit.

Bicycling Magazine and Redfin’s Bike Score also use this flawed ACS data to rank cities.

Open Street Map

Organizations such as People for Bikes and Redfin use Open Street Map (OSM) data to determine bike networks within cities. There are significant problems with that approach in Detroit, including:

  • There isn’t a large, active Open Street Map community in Detroit. This likely true in all cities that have a high poverty rates and low non-mobile Internet access rates. The Detroit data is generally marginal.
  • Detroit’s wide streets and relatively low traffic volumes are very bike friendly, but are not recognized as such by these national groups.

Our Proposal

There is no way of determining the number of bicyclists in a city. However, it is possible to generate an index or dashboard based on multiple data points which could be used to determine trends.
Potential data sources include:

  • Infrastructure miles weighted by their level of separation from vehicle traffic
    • Trails
    • Separated bike lanes
    • Buffered bike lanes
    • Bike lanes
    • Signed routes and neighborhood greenways
  • Bicycle count data
  • Fatalities and serious injury crashes
  • MoGo Bicycle Trips and memberships
  • Slow Roll memberships
  • Number and size of bike racks
  • Social media data for bicycle groups (e.g. Facebook followers)
  • Number of bicycle clubs
  • Number of bicycle shops
  • ACS data
  • Strava data ($, biased demographics)
  • Cell phone travel data ($)
  • DDOT bike rack usage (available?)
  • Google Map directions data (not yet available)

Leave a Comment