Wednesday, June 13, 2012

Car-Free Counties: A Study in Dependent Variables

So I was tweeting yesterday about various TX counties and the percentage of households that are without vehicle, at least according to the 2010 census.  So I've decided to formalize and analyze this data more seriously than a few simple tweets.

First, below you will see a chart of the 6 TX counties measured: Bexar (San Antonio), Dallas (Dallas), Harris (Houston), Travis (Austin), Tarrant (Fort Worth), and Collin (North Dallas, as it is commonly but inaccurately called).  The blue dots represent the Car-Free households per million.  Red shows median income by household.  And Green is Density per square mile multiplied by 10 to get it to register on the graph:

The interesting thing about the TX counties (and I found this to be roughly true for much of the country, at least the car-dependent portions of the country (Sun Belt, Rust Belt, Bible Belt, etc)), is that the strongest relationship between carlessness is not with density, but with income.  It's inverse.  The wealthier the country, the lower the amount of households without cars.  Pretty logical right?  More money means more ability to buy and operate cars.

The r-values of each line substantiates this interpretation with the most noise or statistically insignificant r-value (low) with density.

However, that relationship falls away when comparing with the rest of the country as seen below (and you'll likely want to click to embiggen):

Same chart but with the 67 largest (or larger) counties in the country, mapping the same data.  Red is income, blue is carless, green is density.

You'll notice the noisy data shifts from density to income.  Density and car-lessness is incredibly correlated with r-values very close to 1 (and I apologize, but a persnickety excel file prevented me from including r-values on the graphic).  Income on the other hand goes to negative.  As you can see, income is all over the map in relation to the other two factors.

What this illustrates is that in car-dependent places, going car-less is generally not a choice, but one dictated by income.  Furthermore, because the % going car-free in those places is so low and rare, that means many are even unable to go car-free except for extreme circumstances.  Meaning many have to pay for and maintain a car and those that don't/can't likely face very difficult commutes due to environmental coercion via infrastructure.  In other words, it certainly isn't the most convenient solution, but rather one imposed.

On the other hand, density generally is more of a choice in many cases.  There is minor inverse relationship between income and density, but it is mostly theoretical in that one would expect with more money you can afford bigger space or less money, less space and therefore more cramped or dense living conditions.  However, there are plenty of wealthy people paying for the amenity of living in San Fran or NYC as the income data shows.  This is where theory or conventional wisdom and data differ.

If you notice around the 54th city (x-axis), there is a spike where car-lessness (blue) seems to transition from linear to exponential.  This is Allegheny County (Pittsburgh).  The ones above it are the usual suspects or more walkable places you'd suspect (5 are NYC area including Hudson Co., NJ, San Fran, Suffolk (Boston), DC, Philadelphia).  They're also the cities I b1tch about losing friends to who leave Dallas for more walkable locales.  With what we know about Millennials, these things may very well also be correlated.

Since there is money choosing to live in those dense counties, you can argue that therefore there is greater choice involved in living in both denser places and going car-free.  And I would argue that equates to a far healthier system, city, and market.  And I would also argue that jump represents the difference between the logical, rational, self-ordered city (of past and future), where traffic, proximity, amenity, and quality of place are not mutually exclusive, and the illogical, irrational city of the car-dependent present.