Journey to Work question: are ridehailing apps like uber & lyft considered "taxicab" or "other means" for Table B08301?

Hello, 

I've read the 2018 Subject Definitions document on Means of Transportation to Work over and over again, did a search for "uber", "lyft", "ridehailing", "ridesharing", etc. and nothing comes up.  I have also looked at the Questionnaire Archive, specifically Person Question 31, and it has no mention of these apps.  I am I missing something, or is it really this ambiguous?

Thank you very much,

Diana     

Parents
  • This is an interesting topic indeed. We just finished pulling data for an 8 county region and saw some irregularities in this area that we could only attribute to high mobility. If someone answered that they took public transportation previous to moving to their current address and there is no public transportation where they live now, it often makes it difficult for clients to understand why their regions show residents saying they use public transit.

    I would like to see ACS update their questions to reflect more current transportation trends for sure!
Reply
  • This is an interesting topic indeed. We just finished pulling data for an 8 county region and saw some irregularities in this area that we could only attribute to high mobility. If someone answered that they took public transportation previous to moving to their current address and there is no public transportation where they live now, it often makes it difficult for clients to understand why their regions show residents saying they use public transit.

    I would like to see ACS update their questions to reflect more current transportation trends for sure!
Children
  • I'm on the border of GenX and Millennial, and before reading the doc, I would think it would be either taxicab or other, definitely not driving alone.  Although there is a sentence in the driving alone info about a "passenger serving" type trip in which someone drops you off and then goes back home or to some other non-work location, this is considered driving alone as opposed to carpooling, which makes sense.  (Some examples from my life: way back when, I drove a roommate to her work when she hurt her foot and couldn't drive.  Also, for a while when I was in a one-car household, I was taking the bus to work, but sometimes when it was raining, my husband would drive me, drop me off, and then go back home/go about his day with the car.  I would most likely take the bus home.)  These are totally different from a ridehailing situation (at least in my mind), but it's not clear from the documentation.

    In response to KC_Researcher, It is possible for people to have taken public transportation "last week" even if there's no public transportation in the area if they were on a business trip/working somewhere outside of their regular workplace, or that they are a hypercommuter.  Here's what the Census doc says for that situation (page 94 of the Subject Definitions):

    "The means of transportation data for some areas may show workers using modes of public

    transportation that are not available in those areas (for example, subway or elevated riders in

    a metropolitan area where there is no subway or elevated service). This result is largely due

    to people who worked during the reference week at a location that was different from their

    usual place of work (such as people away from home on business in an area where subway

    service was available), and people who used more than one means of transportation each day

    but whose principal means was unavailable where they lived (for example, residents of

    nonmetropolitan areas who drove to the fringe of a metropolitan area, and took the commuter

    railroad most of the distance to work)."

  • Good points Diana. In our case when we pulled the data from these 8 counties, it was a mixture of very rural counties with zero public transportation and a few counties that are very urban with a fair amount of public transit. So, we chalked off the responses in the rural counties as people who likely moved to those rural areas but had been living in an urban area with public transit. That seemed to be the only logical assumption we could make.
  • Now I'm coming across another wrinkle in all of this...I'm assuming that company shuttles or buses (e.g. Google buses) do not count as "bus or trolley bus" because this is one of the subparts of the "public transportation" macro group. This to me would definitely be "other," right? "Other" would also include scooters/skateboards, and possibly uber & lyfts, depending on how people interpret it.

  • The Google bus example could be "other" but does not fit neatly into any category in my opinion. Since this is not a public transit vehicle, the closest categorization might be as a van ride with multiple people, in other words a carpool.

    This is an area badly in need of updated guidance from the Census Bureau. I hope folks there are reading this thread.
  • This interesting discussion reminds us, once again, of the limits of ACS data, particularly when it comes to journey to work. There is just no way a general-purpose survey, designed to deliver data at very small geographic levels, can provide accurate information on rare (<2%, say) situations. Most people arrive at work in a car, either driving alone or with one or more other people, or take clearly recognized public transportation. Trying to get clarity on ride-sharing vs. taxis vs. some other unusual mode is like it would be to ask one's make of car and expect to get clarity between BMWs and Lexuses, as opposed to between Ford and Chevrolet and Toyota. It's just not possible.

    We really need more education on the inherent limitations of ACS data. The only way to get clarity on mode of transportation is to do a journey to work survey (like the origin-destination surveys of old) and tailor the responses for mode to exactly what's of interest in the metro area in which the survey is being conducted.
  • Personally, I think that it would really benefit ACS to start partnering with other companies that are doing massive data collecting and integrate where it makes sense. For example, Sidewalk labs collects transportation data from Android phones (not sure if Apple is included yet or not), but they also wrap financial analytics into the equation. So, here are some examples.

    Scenario 1: Someone takes 30 minutes to drive say 20 miles. They track that they stayed in the same location for say 2 hours and during those 2 hours they made 1 or more financial transactions. Then, they got back in their vehicle and either drove home or to their next destination. It's a safe assumption to say that this person was either shopping or going to various appointments.

    Scenario 2: Someone takes 30 minutes to drive say 20 miles. They track that they stayed in the same location for 9 hours and maybe they had a financial transaction or maybe they didn't, but it's a logical assumption that this person is there working. Particularly, when they integrate patterns into the analytics.

    Scenario 3: Someone takes 10 minutes to go 1 mile and the route doesn't follow any recognizable roads. There's more to this one than this, but the thought process is that person either biked or ran that distance through a park or unmarked trails.

    Essentially, as I'm sure you already know- it's just basic machine learning (AI). This data has already been collected for a few years now.
  • this will happen, like, never. Nobody is interested in narrow questions like what you presented (except, frankly, for bad actors); matching/record linkage is a task that the researcher needs to performs themselves according to the research questions and tasks they want addressed, while maintaining data confidentiality and the promise that the Census Bureau surveys respondents will never be identified. Remember: micro data can enter Census, but cannot leave it.

    I think the greatest limitation is the speed of adaption / content review. ACS *still* has not recognized the existence of cell phones... the questions on phone use should long have moved from NHIS to ACS, so that survey statisticians like myself could better design ~1,000 phone surveys conducted every year (may be 5,000 a year). (And it is too late now; with all the call blocking apps, the field is moving out of phone surveys to self-completion surveys like mail and web.) Steering the ship that plows through 4M units every year is a difficult operation. Besides, you need to remember that ACS primarily exists to address the *federal* data needs, not your data needs as a researcher or a city planner. Getting anything into ACS that does not have a direct federal data mandate is approximately impossible... unless your Senator can introduce the legislation to put it there and follow through to make it a law.