CASL: A New Data Set with Community College Branch and Satellite Locations
by Dr. Ellie Bruecker, Senior Research Associate, SHSF
When we launched our SHSF Public Transit Map last spring, community college stakeholders regularly asked about branch campuses. Today, students can earn a certificate or a degree without ever stepping foot on a community college’s main campus. And some programs or courses may only be offered at one location, so students need to be able to access multiple locations.
We’re currently building an expanded version of the Transit Map, which will include all satellite and branch campuses for two-year public institutions. To build the expanded map — which will be released later this spring — we had to build our new CASL (Colleges And Satellite Locations) dataset (pronounced “castle”), which is now publicly available.
And oof, I did not expect it to be quite so difficult.
First, just defining the subset of community and technical colleges is tricky. My fellow higher ed researchers, you know this — the predominant degree awarded vs. institution level vs. Carnegie classification battle. IPEDS uses the highest degree awarded to classify institution levels, meaning a community or technical college awarding mostly Associate’s degrees will be classified as a four-year institution. The College Scorecard includes a predominant degree awarded variable that we like better, but it still isn’t perfect!
This is particularly difficult if you’re trying to distinguish between Associate’s degree and non-degree institutions. For example, the University of Arkansas Community College–Batesville offers 17 AA and AAS degree programs and 4 certificate programs. But because of the way the school’s enrollment is distributed and likely variation in graduation rates, it’s classified as predominantly awarding certificates. And while that’s not incorrect, we could debate whether they should be grouped with mostly technical schools instead of community colleges.
Given the hiccups with predominant degree awarded and institution level, a lot of analysts use the Carnegie classification to identify community and technical colleges. But there are 447 public institutions in Scorecard that don’t have Carnegie classifications! Ultimately, we opted to use a combination of identifiers. Our subset includes all institutions that are either Carnegie classified as Associate’s degree-granting OR predominantly award certificates or Associate’s degrees.
After we used these Scorecard variables to define our subset, we matched up these schools with each of their locations represented in PEPS. Then, I decided to do some quality control and check 10 institutions for accuracy in the locations (checking to see if all campuses were included, if addresses were correct, etc). I went to the colleges’ websites and compared the locations they advertise online to the locations in PEPS.
And here was the oof part: I found a handful of missing and incorrect addresses. Not a surprise, but since our intention is to map access to transit, having a dataset with accurate locations is critical.
So, we checked them all. Yep, we visited the websites of almost every community and technical college in America. We corrected addresses that were inaccurate, added locations that weren’t reported in PEPS, and removed locations that we couldn’t verify. Obviously, this process leaves room for human error. We believe that risk is outweighed by the benefits of having a more complete picture of where community and technical colleges serve students.
You can find the data here: https://www.shs.foundation/data. We’ll be updating it soon to include latitude and longitude for each location so users can map the locations without doing the geocoding themselves, as well as some other useful variables.
And while we’re on the topic of maps and data, we also want to amplify some exceptional college mapping efforts we’ve seen recently. In December, my former colleagues at the University of Wisconsin’s SSTAR Lab released a report (and accompanying interactive tool and downloadable data) which used DAPIP data to map college locations in rural areas and categorize the types of locations. And earlier this month, our friends at the Alliance for Research on Regional Colleges developed a new metric to identify rural-serving institutions, creating an interactive map and other data products to support further research. Hats off to both projects, which are huge contributions to the field.
We’d love your feedback and suggestions for how to make the CASL dataset better, so take a look and let us know what you think. And keep an eye out for the first application of the CASL dataset with the new edition of our transit map, coming this spring!
