Did USA Today Publish Misleading Numbers on Bay Area Cycling?

National newspaper writes that bicycle commuting dropped 19.9 percent in San Francisco ... except that it didn't

Cyclists on Market Street
Cyclists on Market Street

Note: GJEL Accident Attorneys regularly sponsors coverage on Streetsblog San Francisco and Streetsblog California. Unless noted in the story, GJEL Accident Attorneys is not consulted for the content or editorial direction of the sponsored content.

“There are three kinds of lies: lies, damned lies, and statistics.”… popularized by Mark Twain.

That quote is used to describe how people often massage statistics to support an idea or to prove a point.

A recent article in USA Today, entitled ‘Fewer Americans bike to work despite new trails, lanes and bicycle share programs‘  suggests that, well, fewer people are commuting to work by bike despite all the efforts to improve infrastructure and make it easier. Although it’s about the country as a whole, one bit of data raised eyebrows in the Bay Area: the claim that from 2016 to 2017 commuter cycling dropped 19.9 percent in San Francisco and 25.8 percent in Oakland.

Here’s the chart they published with the article:

bike_commute_Online

As seen in the caption, the chart attributes the data to the Census Bureau American Community Survey, via the League of American Bicyclists.

So how did USA Today come up with such significant drops?

Chris Woodyard, the writer of the article, confirmed that he got the numbers from the League and not from the source census data. We have an email and a call out to the League.

[UPDATE: The League confirmed (Streetsblog missed an email from them this morning… sorry League!) that they provided the numbers. Streetsblog is trying to find out how they calculated such high drops from 2016-2017]

In the meantime, the San Francisco Bicycle Coalition pointed us to the source data, from the census bureau, which is available online.

Perhaps a math-savvy Streetsblog reader can double check us, but going by the source data, well, there has been a decline in commuter cycling, but it isn’t 19.9 percent or anywhere near it. To get to the original data, go to US Census: https://factfinder.census.gov and then search S.F. or Oakland, and then the commute data is under the “business and industry” tab.

Here are numbers we pulled and crunched directly from the census data:

  • U.S. Census figures for 2017 show the commuting population of San Francisco at 495,315. It shows the cycling rate at 3.9 percent.
  • Multiply 495,315 by 3.9 percent to get the total number of people cycling to and from work. That comes to 19,317.285 people bicycling to work on average in S.F. in 2017.
  • U.S. Census figures for 2016 show the commuting population of San Francisco at 479,139. It shows the cycling rate at 4.1 percent.
  • Multiply 479,139 by 4.1 percent to get the total number of people cycling to work. That comes to 19,644.699 people bicycling to work on average in S.F. in 2016.
  • Okay. The decline from 2016 to 2017 comes to 327.414.
  • Percentage decrease = decrease/the original number × 100.
  • So, take 327.414/19,644.699

…and that decline comes to 1.7 percent.

Maybe USA Today (or the League) was just talking about the decline per capita? That still only comes to about 5 percent for San Francisco. The chart shows an even more impressive decline for Oakland, at 25.8 percent. In reality, the decline was from 3.1 percent of commute trips in Oakland to 2.9 percent from 2016 to 2017.

[UPDATE, Mon. Jan. 7: In a follow-up email, the League explained that the discrepancy is because the figures above are for 5-year *estimates*, and they used a set of numbers for single-year estimates.]

Of course, if cycling is actually down both in real terms and per capita, that’s cause for concern. Some people turn away from cycling, no doubt, because of “…the very real safety concerns that people who bike continue to face on our streets,” wrote the SFBC’s Brian Wiedenmeier, in an email to Streetsblog. Much of the progress in getting bike lanes striped on city streets has been annulled by the rise of Lyft and Uber, whose drivers park with impunity on the bike lanes, rendering them useless. “As tens of thousands of additional Lyft and Uber vehicles continue to flood San Francisco daily, the need for separate, protected infrastructure has never been greater.”

In addition, the census data does not care if your kids walk or bike to school nor that you bike to the grocery store. As such, another possible explanation is that more and more people in the Bay Area are having longer commutes–exacerbated by the housing crisis and housing costs. The census survey “…only asks about the journey to work, and only allows for a single mode answer. So if you biked to the train, your answer would be counted as ‘public transit,’” explained Jeffrey Tumlin, Principal with Nelson\Nygaard Consulting (and the guy who put together Oakland’s Department of Transportation). Also, the census data “…has a high margin of error, making it somewhat dubious for tracking trends of small groups of people in small geographies.”

In other words, despite the assertions in USA Today‘s article, it’s unclear from the survey data if cycling really is down at all. And if it is, it probably isn’t by much.

  • Harald

    You can get close to 19.9% if take the absolute number of people cycling from table B08006 https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_17_1YR_B08006&prodType=table

    2017: 16,266
    2016: 19,429

    (16266-19429)/16266*100 = -19.4%

  • DrunkEngineer

    The census data is useless for gathering trip counts. Doing direct bike count measures gives more accurate picture. The SFMTA gathered quite a lot of data for 2016/2017, which you can view here: https://www.sfmta.com/reports/20162017-monthly-comparison-dashboard.

    In eyeballing the counts, the numbers did seem to decrease by a few percentage points. Note that the sensors don’t detect cyclists outside the facility (perhaps indicates an increasing number of uber/delivery vehicles blocking bike lanes?).

  • Stuart

    If that’s what they did it would be still be a bogus number; a percent change should be measured relative to the starting number, not the ending number, so the denominator in that calculation should be 19429, not 16266.

  • Harald

    Oh yeah, I certainly wasn’t trying to argue that this math would make sense.

  • jonobate

    The correct way to calculate the percentage decrease of the number of SF bike commuters would be 1-(16266/19429) = 16.3%, which is not too far of from the number that USA Today published. The raw number of commuters cycling is so low that any change in the number of commuters cycling is going to look like a huge increase or decrease.

    While USA Today may not technically be wrong, a better way to express this would be “The percentage of commuters cycling to work in San Francisco in 2017 was 4.1%, down 0.2% from 2016.” Always, always ask “percentage of what?” When confronted with stats in percentages.

  • Stuart

    down 0.2% from 2016

    That’s not accurate though; it’s down 0.2 percentage points, not 0.2 percent. You can’t use % as an abbreviation for percentage points.

  • Drew Levitt

    This is why we need to get “basis points” into the general lexicon: “The share of commuters cycling to work in San Francisco in 2017 was 4.1 percent, down 20 basis points from 2016.”

    Good luck though!

  • jonobate

    Sure you can. “percentage points” and “percent” are interchangeable for most people; what matters is that you define what the number is a percentage of. In this case, it’s percentage of total number of commuters, which should be clear from the context.

  • Stuart

    Since your suggested quote doesn’t include the starting number, I don’t see how it’s clear from context that you mean total commuters rather than meaning what the words you wrote actually mean to someone who understands the difference (that the percentage of people cycling was almost unchanged). That’s certainly not how I would read it.

    I agree that that most people probably don’t know the difference between percentage and percentage points, but we’ll have to agree to disagree that it would be better if reporters used them incorrectly on purpose. There are other ways to address that–like giving the starting and ending values, without describing the change directly–that won’t confuse people who do know what the difference is.

  • jonobate

    My full statement was “The percentage of commuters cycling to work in San Francisco in 2017 was 4.1%, down 0.2% from 2016.” I think that makes the context pretty clear. You could change that to “the percentage of total commuters” if you wanted to be more precise.

    I’m not saying that reporters should use deliberately use “percentage” and “percentage points” incorrectly, but I am saying that using those terms correctly is not sufficient without the context of what the value is a percentage of.

  • Jude Williams

    Have any of you thought, even for a second, that maybe people aren’t going to use these other than very occasionally? This has been my issue with these road diets/dedicated bike lanes from the beginning. It’s a nice thought but there just isn’t the demand to justify the impact on the 96% of people for whom these serve no purpose.

    Reading the comments it seems that the people who are for these projects are grasping at straws by saying the drop isn’t that low or by trying to blame TNC cars for blocking sensors. I would expect nothing else than for cars to be blamed though.

  • Stuart

    I think that makes the context pretty clear.

    This is the part we don’t agree on; I did in fact read the whole statement, and I don’t think it’s clear at all. (I’d argue that if you think it means one thing, and I think it unambiguously means a completely different thing, that’s a good signal that it’s not clear.)

  • Joshua Putnam

    The League didn’t just give bogus numbers for the Bay Area, either. They touted the same junk stats for Seattle, where counts of actual cyclists on the streets set new record highs in 2018. https://www.seattlebikeblog.com/2019/01/03/2018-bike-counts-up-32-percent-on-2nd-ave-downtown-after-bike-lane-bike-share-expansion/

  • George Joseph Lane

    It’s important to note how the question is asked as well. I used to work in New Zealand and the census numbers were next to useless as the question was ‘What was your main mode of travel to work on the census day’.
    New Zealand’s two main cities are wetter than Seattle and windier than Chicago so wet and windy days see almost 50% less cycling than the average day. The census didn’t count people who cycled to a transit station, or to school/university, or for other utility trips. As a result, Auckland installed about 20 permanent cycle count stations to try and capture accurate annual data to show the changes throughout the city.
    Auckland’s permanent counts have been useful in other ways too; for example, by showing that building a new cycle path parallel and half a mile away from an existing one actually increased cycling on the existing one by more than the regional average increase. This has helped to dispel the myth that new infra only attracts cyclists from other routes.

  • sf in sf

    While I do see “down X%” used as a shorthand for “down by X percentage points” fairly often, it’s not correct, and I don’t think it’s great to deliberately choose that presentation to paint a misleadingly rosy picture.

    Let’s face it, commute cycling is significantly down. We gotta figure out why and reverse the trend. It does no good lying to ourselves.

  • linked1

    There was a real sense of optimism before 2016, and cycling culture was having a moment – that almost became a movement. But then Hillary stole the nominations from Bernie and the US descending into the dark ages. Now cyclists are afraid to ride in the streets for fear of being run down by coal-rolling monster trucks brandishing ‘don’t tread on me’ flags.

  • The author (and apparently the San Francisco Bicycle Coalition) is using the wrong census data.

    The American Community Survey comes in multiple period estimates: 1-year, 5-year, etc. The 3.9% number comes from the 5-YEAR ESTIMATE for 2017 which is the period from 1/1/2013 to 12/31/2017. Comparing this to the 2016 5-year estimate (i.e. 1/1/2012-12/31/2016) is just plain wrong.

    You should be looking at the ACS 1-YEAR ESTIMATES if you want to get the 1 year change. Here’s the correct data:
    2017 16,266 <- down 18.5% (16266/19949-1)
    2016 19,949 <- down 6.9%
    2015 21,427

    Correct Census data source:
    https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_17_1YR_B08006&prodType=table

    Info on ACS period estimates:
    https://www.psc.isr.umich.edu/dis/acs/handouts/Compass_Appendix.pdf

ALSO ON STREETSBLOG

A rendering of the bridge bike and ped path. Funding is now available to study turning this into another car lane. Image: MTC/HTNB

Marin Supervisor Asks Officials to Study Converting Future San Rafael Bridge Bike and Ped Path into Another Car Lane

|
Note: GJEL Accident Attorneys regularly sponsors coverage on Streetsblog San Francisco and Streetsblog California. Unless noted in the story, GJEL Accident Attorneys is not consulted for the content or editorial direction of the sponsored content. A Marin County Supervisor, who represents San Rafael, apparently wants to look at nixing plans for the long-planned bike and ped path on the […]