Sunday, November 9, 2014

Public Transit: All About Density

I grew up outside of a small town of about 8,000. If we wanted to go anywhere other than the local crick we needed to drive. Imagine my surprise when I was 14 and visited Montreal: a massive subway system and underground malls! Big city folk had a whole different lifestyle.

Now I live in Atlanta and drive everywhere.

My experience in Atlanta isn't unique. See the graph below, which shows average daily usage of all public transit in a city. (You can filter to a specific type in the bottom right.) New York City has more than twice the per capita public transit usage of any other metro area. Most metros have less than 0.1 daily trips per person. Those trips are unlinked (i.e. a transfer counts as another trip) and don't count return trips, so in most metros less than 5% of the population uses public transit on a given day.

This leads people to say, "the public transportation in city X sucks". We've all heard it. But look at the graph again. New York City has twice the transit usage of any other city, but is also more than twice as dense (denoted by color). Why does this matter? Based on average density, in New York City 6,100 people live within a quarter mile of each metro stop1 while 430 people in Atlanta live within a quarter mile of each metro stop. Also, the alternative (driving) is much worse in a denser city. Traffic may not be great in Atlanta, but its far worse in a city with 15 times the density.

So, how does Atlanta's transit stack up when controlling for density? See the graph below. Atlanta is one of the least dense big metros in America, but compared to similarly dense metro ares, Atlanta has a high rate of public transit usage. For example, observe Jacksonville which also has a little more than two thousand persons per square mile. Jacksonville has about 0.03 daily trips per capita and is right on the trend line (the average for similarly dense cities). Atlanta has over twice the usage of Jacksonville.

This visual is full of other fun information. Washington DC and Boston are the two biggest public transportation over-achievers. Los Angeles, even though it has more public transit than most metros, is one of the biggest under-achievers, along with San Jose and Riverside. Click on the Los Angeles circle to see that nearly all of their public transit comes from bus, or click on the orange segment of the US map to see that western cities fall into two distinct trends of public transit vs. density.

The last visual gives us an idea of how a city's transit use is changing. Raleigh has more than doubled their transit use in the past twelve years, but their current rate (as represented by color-coding) is only 0.02 per capita, so the actual increase was slight. Scrolling down the graphs and looking for color outliers shows that Salt Lake City, Seattle, and New York City all had impressive gains in metro use.

Atlanta's public transit use declined by 9% over the past twelve years and was the only city to see a decline in heavy rail usage. Where does this put Atlanta going forward?

This post is about public transit, but what really matters is quality of life. I don't actually drive everywhere; I live in the Old Fourth Ward and can walk to great coffee shops and parks. I have a ten minute driving commute to my office downtown. Atlanta's Beltline project and related push for mixed use development should make a lifestyle of walkable neighborhoods and short commutes accessible for more people. And as that happens I expect Atlanta to grow in density and public transit.

  • All population information is from the 2010 census, and uses population-weighted density to be more meaningful. 
  • Transportation data is from the American Public Transit Association, available on their website. (See the monthly data)
  • The APTA uses urbanized areas (UZAs) to report usage, while I only had population-weighted densities for metro areas, which sometimes encompass multiple UZAs. I was able to merge the two with a name crosswalk, but did have to look up a few by hand.
  • I suspect a reporting error in the 2002 APTA data for Austin. Their drop is entirely from 2002 to 2003, and in privately provided public busing.

1 Calculations are totally based entirely on average population-weighted density, and not on actual stop locations.

Charter School Performance in New York City

A couple months ago I wrote a few posts with my friend Jarod Apperson using New York City public school data: 

Success Academy Posts:
Part 1 * Part 2 * Part 3

Here's a preview of my favorite graph from those posts, which shows Success Academy schools performing far better than their economically similar peer schools, really an astounding amount:

We also wrote this follow-up, which provides some context for some of the other charter systems: Performance by Starting Grade

And we converted the Success Academy posts into a Sunday op-ed for the New York Daily News.