Visualizing Durham Technical Community College
Service Areas

Exploring the use of census data in making choropleth maps.
David Hopp, Ph.D.

This is an exercise in the use of census data and how it might provide insight into characterizing the service areas of North Carolina community colleges. The kernel question is, "where do people who attend a community college live?" Asking that question begins a cascade of related issues. We will look at a few of these here without any attempt to be comprehensive.

A first consideration might be to realize that, even if we knew the street address of every student (we wouldn't be able to use that data directly because it is personally identifiable information (PII) which comes under the scope of FERPA, but even if we could) what would we do with those tens of thousands of items? Well, plot them on a map sounds like a start. But what good would thousands of points on a map be? Sounds like a census, doesn't it, and censuses roll-up location data. Besides that, censuses provide all kinds of useful information, like age, gender, family size, income, etc. So it would no doubt be to our advantage to see our service area information in conformity with census information.

Ok, but a census is only every ten years and things can change quickly, especially in urban areas. Aha, censuses used to be every ten years but as of about 2005 the Census Bureau uses its American Community Survey (ACS). The consequence is more timely data, although with more statistical uncertainties. Combined with that are the Zip Code Tabulation Areas (ZCTA), in use since the 2000 census.

The ZCTAs are a big help to us in our project. They roll-up ZIP codes (odd beasts, they belong to the Postal Service and can change without notice) and are for the most part comfortable geographical regions - that's good for map-based visualizations. Add to that the variety of quantitative data associated with them and we have a great fit for our service areas. All we need from a student data base is ZIP code (not a PII), and some elementary demographics, like gender and age (still not PII). In the community college context we would qualify these records with the category of classes students take - developmental, curriculum, and continuing education. That should do it.

If we are going to talk about maps, we have to be more specific. We have decided to use choropleth maps that display ZCTAs and county boundaries. "Choropleth" is a word of somewhat recent vintage. Its two parts are derived from the Greek. "Choro" has been in use for a long time (in our case it has nothing to do with a genre of Brazilian music, nor with a choreographer, who produces choreographies) in describing the art or technique of mapping regions. "Pleth" is something you may have encountered as a plethysmograph in a doctor's office. The word refers to quantity, rate, or relative proportion. A choropleth map indicates how much of, or to what degree, something exists. Practically speaking, a choropleth is a map colored to show regional concentrations of whatever you might be interested in.We will stay out of the arguments that demographers get into about what "really" is suited to a choropleth map and proceed in innocence. Here is an example of such a map.

We do not at present have the student-ZIP code information, but we do have publicly available census information directly from the Census Bureau. We show below three choropleth maps for the Durham, North Carolina, area. The maps show Durham and adjacent counties, namely, Wake, Chatham,Orange, Person, and Granville. We used the R programming language to produce the charts, and in particular the choroplethrZip package for R. R is an open source programming language, maintained by a world-wide community and used extensively in statistics, genetics, finance, and many other areas.

All three of these maps show the same region but with slightly different demographic information. Their most salient features, however, are the locations of two Durham Tech campuses and of three Wake Tech campuses. The main campuses are shown as circles and satellites as triangles. Each campus is at the center of a 10 mile circle, its putative service area. Until we get more student data the 10 mile radius is somewhat arbitrary, but not unreasonable. The circles appear more like ellipses, but that is due to mapping considerations - the world is not flat no matter that pages and computer screens are. For those readers interested in navigation, the circles using the rhumb, not the great circle, distance. Rhumb lines (loxodromes) follow a fixed bearing and correspond to the idea of walking a straight line on a Mercator projection. Be that as it may, we are speaking only of short distances compared to the size of the Earth so all this makes little difference except to geographers and others concerned with what is called spatial data.

The first map shows total population. The second shows population age 18 through 24, while the third shows age 25 through 49.

                                Total Population                        18-24 Years Old                          25-49 Years Old

Updated 2015-05-20