In Friday’s excellent League of American Bicyclists webinar discussing “Equity in Bicycle Advocacy and the ‘invisible cyclist'” (storified nicely here), one topic of discussion was the challenge of counting bicyclists. Although there seemed to be general consensus that the term “invisible cyclist” is either inappropriate and offensive:
There was also plenty of attention to the reality that our methods of counting bicyclists systematically overlooks certain populations of people using bikes as transportation and the complexity of their reasons for, and experiences during, bicycling:
I happen to be teaching a course right now called “Methods and Approaches in Environmental Studies” in which we’ve been discussing certain inherent limitations to quantifying social life. As if this isn’t problematic enough, in environmental fields our goal is often to understand the dynamic interrelationships of human and natural systems. When the human social reality can’t be quantified, we often employ qualitative methodological approaches. But this introduces the challenge of integrating qualitative data with the quantitative data typically derived when measuring physical variables like environmental change.
The same challenge is playing out in the bike advocacy movement. Counts of bicyclists seem to be essential in justifying expenditures on infrastructure to support bicycling. But the ways we count generally fail in two fundamental ways.
First, when counts are based on field observation methods, they are generally based on limited and non-random samples of the hypothetical population of existing cyclists. Over on Invisible Cyclist Julian Agyeman and I have written about this problem here and summarized the argument again here. The problem with counting bicyclists at the places and times where there seem to be a lot of them should be obvious. Even a strategy of conducting bike counts at randomly chosen locations and times, while at least systematic in terms of “event sampling,” can tell us about frequencies of bicyclists passing through randomly chosen points in space and time but not about who those cyclists are, where they began their trip, where it will end, how frequently they make the trip, or what the factors were shaping their mode choice.
Another approach is to randomly sample from the general population and then ask subjects to self-report their transportation choices (capturing what usually looks like a very small amount of bicycling as transportation). The best example of this approach is the U.S. Census Bureau’s American Community Survey (ACS). The ACS asks of the person responding “How did this person usually get to work LAST WEEK?” and instructs respondents who used more than one mode of transportation (e.g., biking to a transit stop), to mark the mode used for the longest part of the trip. The Census Bureau’s own analysis of the ACS acknowledges how this wording excludes many possible trips by bike:
Because bicycling and walking often serve as secondary travel modes that supplement modes such as transit or driving, some commutes that involve bicycling and walking are not reflected as such in the ACS because another mode is used for a longer distance. (see Modes Less Traveled—Bicycling and Walking to Work in the United States: 2008–2012 [PDF])
This is the first limitation in the ACS data. The second is that it is asking exclusively about commuting. According to the Bureau: “The analysis is limited to workers 16 years and over who worked during the ACS reference week, the calendar week preceding the date respondents completed their questionnaire, and who did not work at home.” This is problematic given that work commutes are declining as a percentage of the total number of miles the average person travels. As the American Association of State Highway and Transportation Officials reports in “The Role of Commuting in Overall Travel” (PDF), in 2013 just 16% of all person trips and 19% of all person miles of travel were work commutes. This is down from a high of 40% when first measured in 1956. Meanwhile, the types of trips that are on the rise include errands, family/personal-related, and sports/recreation-reated trips. (Other limitations in the ACS data are summarized in the Bike League’s #bikedata twitter conversation, storified here).
All of this brings me back to two very important questions that I constantly urge my students to consider: How does your research question shape the variables to be measured? And how are those variables going to be operationalized (i.e., what are the specific steps you are going to take to observe/count the phenomena of interest)?
So I would encourage the community of bike advocates and activists, not to mention urban planners and public officials, to be clear about what we are asking.
- Do we want to know who the people are who are currently riding bikes as transportation? Or do we simply want to know the numbers of people riding?
- If the former, do we want to know the bicycle user’s perceived safety and the limits to perceived safety beyond which she will choose another mode? Do we want to know how certain types of infrastructure influence perceived safety? If so, among which individuals–those who are already cycling or those who are not?
- Do we want to know how mode choices are made? Do we want to know about when multi-modal trips, where the bike is at least one mode, are taken?
Note that most of these questions lend themselves to either quantitative “counts” or qualitative “accounts.” Finally, all of these questions need a spatial dimension, which is where issues of equity and justice enter. Do we care about the answers to these questions, and how they might differ, across all urban spaces? Or only in certain places (e.g., where cyclists already turn out in large numbers)?
Until we become clearer about the questions we want to ask about bicycles and those who ride them, it will be difficult to design methods to answer our questions. And the questions we need to ask will depend heavily on what our goals are. The goals of urban sustainability, defined strictly along lines of carbon footprint, may require different questions than public health-related goals or economic development goals.
Bicycling is still seen as a niche occupied by an increasingly powerful interest group with the narrow goal of increasing public expenditures on infrastructure and amenities to benefit the 2% of Americans that our surveys tell us are commuting by bike. Yet carbon footprints, public health, and economic development are obviously inextricably linked, each making equally significant and interrelated contributions to the common good. If we frame our work as bike advocates in terms of the broad goal of contributing to the common good, what would be the kinds of questions we would need to ask about bicycling? From the answer to this question, I guarantee we can put our heads together to identify innovative approaches to “counting” bicyclists that also “account” for the rich diversity of experiences among those who ride bikes.