Archive | Transit Equity

Our First Transit Equity Results

Patience, there’s going to be a little bit of detail here…

We have completed our first full (and raw and tentative) data set for the equity project. We have accumulated transit scores for over 12,000 points in the three-county area.

I decided for the sake of completeness to include all census tracts that include some portion of the TriMet service district. This is includes a tract that has a few bus stops in Estacada and a whole bunch of the Mt. Hood National Forest. There are several of these large, mostly outside, tracts, so in fact about a quarter of the sampled points have a zero transit score.

This should get sorted out when we shift to block group level data.

But we now have a complete map (yes, there is indeed a donut hole in the service district near Happy Valley):

A more useful map is this browsable one (I’m learning about the ways to load KML, but don’t understand them all yet!). If you click on a tract, you’ll get a bubble with the tract ID and the score. Click on the tract ID and you’ll launch a new map with just that tract and our sample points for the tract. You can click on those to get the scores at those points.

So that you can fully look over our shoulder, all the data for this data set (I’m calling it our October 2010 set) can be found here:

In addition to the Census shapefiles we used, you’ll also find these items:

Tomorrow In a few days we’ll start looking at matching this up with some demographics.

What’s the Right Scale for Our Equity Analysis?

The first pass of scoring is complete, sort of. The map of all the selected tracts is below, and it’s apparent that the strategy of selecting just those tracts that have bus stops is insufficient, there are a number of census tracts inside the TriMet service boundary that have no bus stops. So I need to do a bit more work on that.

But I’m more interested in the fact that the map is pretty “flat”. There aren’t a lot of variations from tract to tract (there is of course the expected increase in scores as you move toward the center of the region). I’m wondering if a census tract is too large a unit, if it “evens out” differences in service levels across too big a geographic area.

One commenter suggested using census block groups, the next smaller unit (about 3 or 4 per tract). I’m going to look at that because the ACS data will have some information at that level.

So look for another map, hopefully by next week. I’d welcome other thoughts or suggestions. And I’ll get a tabular data set of points up once I’ve filled in the holes in my map.

What Data Set to Use for Our Transit Equity Project?

What demographic data sets will we use to compare against Transit Score to assess equity?

The 2000 Census is currently the only complete data set at the census tract level, so we’re going to start there. Obviously that’s way out of date and we shouldn’t reach any strong conclusions based on it.

I was asked in an earlier thread if ACS (American Community Survey) data was available at the tract level and I answered that it was. I was almost right. ACS is based on annual sampling (as opposed to the complete census that happens every 10 years) but features a longer questionnaire and therefore more data items.

In order to accumulate enough data points for areas as small as a census tract (a census tract has about 8,000 people, give or take several thousand), it takes five years of ACS sampling. ACS will release its first 5-year data set in December 2010, so at the point we would switch over and have much more current data. The 5-year data sets will then be updated annually (kind of a moving average) and that would likely be the preferred data set on an ongoing basis.

And on the mapping front, with the help of a KML file from TriMet, we’ve added the outline of the service district to the map. Our scoring process has finished Clackamas County and has now moved on into Multnomah County.

Filling in Our Equity Map

Following up on our Transit Equity post yesterday, we’ve begun grabbing our Transit Scores. It’s going to take about 10 days to fill them all in.

But meanwhile we’ll show you the work in progress. This map has the census tracts that are complete so far, with the color representing the average Transit Score for the tract (hotter color = better transit). You can watch this fill out over the next few days!

I’m also going to try to generate more of a heat map display (which won’t worry about the tract boundaries) – watch for that later…

You can also expect to see our data in tabular form soon as well.

Our Transit Equity Project

It will be no surprise to our readers to hear that that there are some in the community who believe that our transit system has become less equitable in recent years as light-rail openings and budget-driven bus cutbacks create a perception of different service for different parts of the community.

But is that accurate? Light rail certainly does not exclusively serve affluent neighborhoods. Can we find a way to get past the anecdotes and accusations and actually quantify who’s getting served how well by TriMet?

We have a new tool to help with this. As we discussed last month, Transit Score is now available. It’s a new tool, and it may not be perfect, but it gives us an easy-to-comprehend number that’s objective (at least it’s not derived by anyone with a view about transit operations here in our region).

So how do we propose to make use of Transit Score to answer our equity question? We plan to:

1) Aggregate a Transit Score for every census tract in the TriMet service area (there are 286!).

2) Correlate those tract-level scores with other information about the census tracts like income, ethnicity, density and potentially other factors.

We’ll do this whole process in an open way, so anyone who’s interested can look over our shoulders and verify our data, or come up with their own alternative analysis.

The rest of this post will be about how we’re accomplishing step one – coming up with an aggregated transit score for a census tract.

The approach we’re using is to lay down a grid (about 1/4 mile) across each census tract, then use a geocoding service to snap each grid point to the nearest intersection. This keeps us from calculating a transit score for a corn field, and also eliminates redundancy in less dense parts of the street network. Depending on the census tract, we wind up with a half dozen to a couple of hundred points in the tract. An example of such a set of points is shown below.

In the next step, we’ll get Transit Scores for each of those points – we have over 11,000 to get – and compute averages for each tract. More about that, and how you can help, tomorrow…