Transit Equity: Starting to Look at ACS Data


You’ll recall that back in October we did our first correlation between Transit Score and several demographic factors, based on the 2000 Census. We came up with the following correlation coefficients:

  • Transit Score/Density: 0.67 (1.00 would be perfect correlation)
  • Transit Score/Percent non-white: 0.40
  • Transit Score/Median 1999 Household Income: -0.52

Based on that, we set two key goals for our next steps:

  1. Get newer data!
  2. Look at finer-grained information, specifically at the Census “block group” level rather than the Census tracts we had used for our first analysis

We went on to compute Transit Scores at the block group level in November.

On December 14th, while we were busy getting ready for the holidays (and playing with Transit Appliances) the Census released the first set of American Community Survey (ACS) 5-year data.

ACS is important because it is updated continuously (not just every 10 years) and because it contains much more demographic information than collected in the Census. But it is a sampling process, not a census, so the data is an estimate.

The 5-year data represents an average of ACS data collected between 2005 and 2009, and is the first data set that gives results down to the block group level. So over the break between Christmas and New Years I grabbed the data and matched it up with our Transit Scores. You can find the correlated spreadsheet here (MS Excel). And here are the correlation coefficients:

  • Transit Score/Density: 0.53
  • Transit Score/Percent non-white: 0.23
  • Transit Score/Median Household Income: -0.39

Obviously the first take is that the correlations are weaker than they were for the prior data set. But the reasons need a little bit of looking into. Tomorrow we’ll explore some of the possible factors…


One response to “Transit Equity: Starting to Look at ACS Data”

  1. One possible reason for the weaker correlation might be the abysmal margin of error rates (+/- 60%+ for some blockgroups) due to the low sampling of the ACS.

Leave a Reply to Matthew Cancel reply

Your email address will not be published. Required fields are marked *