How far do people walk to take transit? This is a classic question in transit planning, and the most common answer is “400m”, represented as a 400m circle around the stop. A longer distance, 800m is also often cited, particularly around rail stations. (In the US, these round very neatly to 1/4 mile and 1/2 mile.) What actually happens out there?
I looked at the commute mode reported in the 2016 census. There are some important caveats — this is only the mode that workers use to commute, and thus doesn’t reflect other users of the system, like students going to school or people travelling for errands. It’s also the self-described “usual mode of commute”, which can be biased. With that being said, it’s also one of the best data sources available, especially since it’s comparable in Canada and other countries.
The smallest geography with good coverage from Statistics Canada is the Dissemination Area, (DA), which covers around 400 to 700 people. Here’s the DAs in Calgary, along with their walking distance to LRT stations:
I don’t think this is surprising; obviously the closer you are to a station, the shorter the walk distance. Here’s the transit mode share:
It’s pretty clear that people who live near transit stations take more transit. One thing that’s interesting is that there is a relatively low mode share for transit in the inner city — what’s up with that? This:
This is the “active mode” (read walk + bike) mode share. The inner city isn’t taking transit, because they’re within walking distance of their jobs — and paying a premium to do so. There’s also other places, like the northwest around the U of C and Foothills hospital, and even near Chinook Centre, where a lot of people can walk to work. It’s not really fair to expect transit to compete against walking, so I think this is the better view:
This is the transit share of motorized modes; that is transit / (transit + car). It represents the choices of people who are travelling beyond walk distance, which is really transit’s core market. And the trend is much clearer for high transit use around the stations. Another way to look at this is through a plot:
There’s a lot of noise — DAs are a small geography, and there’s a lot going on — but also a clear signal, where the closest DAs to transit have substantially increased transit commute share. (This is also an artefact of the nature of DAs — the closest ones to transit are mostly relatively small, and since they are drawn on a population basis, that means they are likely to be relatively dense. In fact, most of the DAs within 300m of LRT are in the Downtown West Village area of Calgary, an area entirely made of highrise apartments.)
The trend is clearer in a version of this, where the DAs have been binned based on their distance to transit. I’ve binned the closest 1% (about 9 DAs), the next closest 1% and so on up to 10%, then gone by 2.5% intervals (22 or 23 DAs) to 30% and 5% intervals beyond (45 DAs per bin), considering all the DAs up to the 3.5 km mark.
Those first couple of bins have really high transit usage for Calgary — the closest bin has an average of over 50% motorized commuters choosing transit. What’s interesting, though, is that the trend continues past 400m, and even to a lesser degree past 800m — it’s only in the 1600m or so range that the curve really flattens.
One other way of binning is by walking distance band:
The under 400m band — which I remind you is disproportionally in the west downtown — really stands out here, but you can see a drop off all the way to 1.4 km or so. And that 400-600m band is still 70% more likely to take transit than the distant residents; the 600-800m is 50% more likely.
I think it’s a useful reminder of a few things:
- Walking distance is important!
- A 400m walking distance is too small for a frequent rail service, and an LRT station may have an effect up to 1200+ m. Note that these are walking distances, not straight line distances.
- There’s a real nonlinear effect; going from a few hundred metres to 600 or 800 could half the transit use.
This last point is really important when thinking about true transit oriented development — it’s not just shoving a bunch of stuff within a magic 800m circle of a station, there’s a real importance to the land closest to the station. Right now, an awful lot of that land in Calgary is park and ride lots.
I’ve skimmed over some details about how the data was processed and what makes up the “poor quality” DAs that were removed; these are in a separate methodology post here.
Mean Median Mode Choice 