Did Chicago Bike Commuting Really Dip in 2013?

Screenshot 2014-09-18 14.46.52

The Census Bureau released American Community Survey estimates for 2013 last week, which report that the number of people biking to work in Chicago has decreased from 1.6 percent in 2012 to 1.4 percent in 2013. This could well prove to be a one-year blip against a broader, multi-year trend that has seen bike commuting nearly triple since 2000.

Other cities have also seen temporary declines in this key statistic, then went on to see strong gains. In New York City, the Census reported that the proportion of workers commuting by bike dropped in 2008 and 2009, but then doubled in subsequent years. Streetsblog Network’s earlier story on the data release included a chart from Bike Portland showing the growth of bike commutes in various cities, with substantial year-to-year fluctuation across all of them. For instance, Portland’s commute rate dipped in 2007, then rocketed by 50 percent in 2008 — and that gain has stuck ever since.

It’s easy to overstate the importance of the Census’ bike-to-work statistic. This number is widely reported mostly because the reliable Census Bureau updates it annually, and makes it easy to find and compare. Ultimately, though, national advocates acknowledge that it’s a woefully incomplete measure of bicycling, and not just because it only measures commute trips.

The particular dataset released last week, the American Community Survey one-year estimate for 2013, is particularly unreliable because it draws from a very small sample: The total number of Chicagoans who told the Census they biked to work last year numbers around 400. A shift of just a few dozen people, for reasons like poor weather or pure chance, can impact the final tally. Due to the small sample size, the survey has a margin of error of plus or minus 0.2 percentage points. So, the true mode share in 2013 could have gone up relative to 2012, decreased even more, or stayed the same.

The American Community Survey also reports estimates based on several years of surveys, which are more reliable but which would be even slower to pick up on rapidly changing phenomena. Bicycling in Chicago is improving quickly, and even the 2013 survey (which is administered monthly) wouldn’t have captured the full impact of many miles of new protected or buffered bike lanes and a huge bike-sharing system — both of which should meaningfully increase the number of Chicagoans biking to work.

The more reliable multi-year estimates from the American Community Survey are due later this year, and in 2015 a new National Household Travel Survey will give us a better look at bicycling for non-work trips. Chicago doesn’t have to wait for the feds to tell us about what’s happening on our own streets, though. The area can always do its own measures of bicycling, for instance through in-depth travel surveys (perhaps piggybacking on the NHTS, as other areas do) or through on-street counts.

Updated with a clarification about the survey’s margin of error.

  • Lisa Curcio

    I don’t pretend to understand anything about the data, but do the bike counts that CDOT performs give us data that is comparable?

  • I believe “no, CDOT’s bike counts give no comparable data”. That’s not to say that the bike counts aren’t useful, but they are location-specific counts, and not a count of Chicagoans who bike (to work).

    For example, CDOT conducts myriad count types:
    1. Project-specific counts. This includes counting vehicles, bicyclists, and pedestrians in a corridor or at intersections to prepare for a project, say Division/Clybourn. The counts will happen for a couple hours usually in a single month. They’re not published online.

    2. Cordon counts. These are done four times a year at certain “entry” and “exit” points to the Loop and River North, but aren’t done at *all* of the points. New York City, on the other hand, conducts the same count at all entry and exit points to Manhattan across 50th Street, and all bridges, since 1985.

    3. Same intersection counts. Every month CDOT counts the number of bicyclists for 4 hours on one day at month at the same six intersections.

  • skyrefuge

    The entire premise of this blog entry is wrong. The ACS did NOT show a drop in the number of people biking to work between 2012 and 2013. Margin-of-error is an actual thing. It’s not something to just ignore and throw away. Any time you write a mode share percentage without a margin-of-error, you are misquoting the ACS, and saying something that they never actually said. The actual numbers reported by the Census Bureau are:

    2012: 1.6% +/- 0.2%
    2013: 1.4% +/- 0.2%

    So maybe the “real” number in 2012 was 1.4%, and in 2013, the “real” number is 1.6%.

    Things the ACS did NOT say:
    – Bike-share decreased 0.2 percentage points from 2012 to 2013

    Things the ACS DID say:
    – Bike-share MIGHT have increased 0.2 percentage points from 2012 to 2013
    – Bike-share MIGHT have stayed the same from 2012 to 2013
    – Bike-share MIGHT have decreased 0.6 percentage points from 2012 to 2013

    A rougher way of saying all this is that the ACS data is “noisy” (which the ACS fully admits), and it’s a waste of time to take anything from it based on any single year’s report.

  • Matt F

    prob bc i moved to Seattle. j/k

  • Thanks for the feedback. The change is certainly within the margin of error, and that should have been clearer. That’s more or less what we were trying to say in the middle of the article, by citing the small sample size and the unreliability that follows from that.

  • I’ve updated the post to mention margins of error.

  • skyrefuge

    LOL, as your post showed up I was writing a response to Payton to remind that as this is a web-publication, it can be updated. :-) Thanks. I think the first sentence (and the graph) are still factually incorrect, but it’s an improvement.

    The blindness towards margin-of-error affected the wider StreetsblogUSA/BikePortland narrative too. Looking at more cities reveals that Chicago’s margin-of-error is really quite small relative to other cities. NYC’s is also small, so city size might be a factor. The narrative was that the bigger/more-established biking cities are seeing slowing/declining growth, but that might really just be a reflection that the numbers for those cities are more accurate and less noisy.

    Most other cities have something like +/-0.7 percentage points for their margin-of-error. With that kind of uncertainty, it will be easy to pick out cities in the noisy data-set that show big year-to-year “gains”. Tucson, New Orleans, and Pittsburgh were singled out, but really that’s not much different than singling out cities that got “heads” each time when they flipped a coin three times in a row. It would be just as easy to cherry-pick cities big “losses” (cities that got “tails” 3 times in a row) if that’s what you were looking for.

    For example, Rockford “plummeted” from 0.7% bike-share to 0.3%. A 57% drop in one year!! But the real numbers are:

    2012: 0.7% +/- 0.6%
    2013: 0.3% +/- 0.3%

    In other words, ACS is saying “we have no idea what the bike-share in Rockford is. Though we’re more confident in our 0.3% number than our 0.7% number!”



Census: American Bike Commuting Up Nine Percent in 2012

Congratulations, America. We’re biking to work more than ever before. We’ve known for a while that Americans are driving less than they used to, even as the economy grows. And just about every quarter, the American Public Transportation Association delivers more stats about increasing transit ridership. Now the Census brings another measure of Americans’ shifting […]

Bike Commute Rate in Portland Reaches a New High

New Census data out this week shows that the bike commute rate in Portland, is higher than ever, exceeding the 7 percent threshold for the first time. Meanwhile, in the tier below Portland, about half a dozen large and mid-sized cities are neck and neck, Tom Fucoloro at Seattle Bike Blog reports: Seattle (3.7 percent) is now in a […]