Directions Magazine learned about RegionalDifferences.com
from Travis Nels on Twitter. The site invites visitors to draw the boundaries of named regions in the United States on a Google Map. After a bit of digging, we found the person behind the site, Clarence “Sparr” Risher. He shares what prompted the development of the site and his plans for the data he’s collecting.
Directions Magazine (DM): The website is aimed at capturing individuals’ definitions of regions in the United States. Why did you want to capture these data? What will you do with the data? Will the results be shared publicly?
Clarence “Sparr” Risher (Sparr): Through my childhood and adult life I have lived and worked in many different places in the eastern half of the U.S., Ohio and Texas and Florida and every state (literally) in between. This has led to encountering a lot of people with very different ideas about the subject matter of this project. Attempting to discuss the topic usually leads to incredulity on their part, and bafflement at the idea that people from other regions don't have the same regional definitions. The biggest disagreements that I encountered prior to moving to Massachusetts were regarding the "Midwest" and the "South,” however those do not hold a candle to the outrage of people from the Northeast when confronted with alternate definitions of "New England.”It was that last one that finally inspired me to start this project after having conceived of it much earlier.
I don't have any serious plans for how to use the data. I am enjoying using preliminary results to educate people on this matter. The results will certainly be shared publicly. They already are, actually. Anyone with knowledge of Web services can automatically request the data from the service that feeds the interactive graphical results map shown to participants after they submit their data. I hope these data are useful to others in linguistic and cultural research in the future.
DM: How much data do you hope to collect? Are you looking for a representative sample of “home” regions?
Sparr: I've already collected about 700 responses, which is more than I originally hoped for and more than I need to see basic trends. I do want broadly representative samples of "home" regions, which will allow for more nuanced analysis of the data. The project on which mine is based (see below) published its results after a few months and then again after a year, but it never stopped collecting data. Lacking any strong motivation otherwise, I'll follow that plan somewhat, and I would expect to have a few thousand responses in a year.
Sparr: I do not have a background in mapping or GIS. I do, however, enjoy data collection and visualization, and am a fan of interesting maps and infographics. I did not build the platform for this project from scratch; it is actually a direct fork of the code used by Andy Woodruff's team for their Boston neighborhood survey, which they published on GitHub and has been used in a few other smaller projects already, such as the Neighborhoods of Burlington, VT. My fork of the platform uses Google Maps API with the Drawing library for the input, stores a variation of GeoJSON data in its database, and utilizes Mapbox.js for the results output. I will be publishing my code back to GitHub when I am comfortable with its stability and usefulness, and I hope it is useful to others with similar projects in the future.
DM: Visitors are asked to draw, or skip if they don’t know the boundary, 27 specific regions. Where did that list come from? What might you learn from the regions that are skipped?
Sparr: After multiple attempts to find lists of regions, they all turned out to be subsets of a list from Wikipedia, which was the basis of my final list. From there I did not keep regions that were numbered or identified by a city (rather than having their own name), regions with hard practical or legal boundaries (court circuits, time zones, etc.), regions within a single state, or regions covering a relatively small land area or population. That produced a list of approximately 40 regions, which I thought would be too many on which to expect most people to provide data. I then applied my own judgment in narrowing further, eliminating regions that I expected to have the least variance or be recognizable to too few people to be useful in the survey.
The only thought I have so far in terms of skipped regions is the possibility of eliminating the most often skipped regions from a later version of the survey, but I am unlikely to do that. I suspect other data analysts will have better ideas about utilizing that information. Generally speaking, I expect professional statisticians and mapmakers to put my data to much better use than I can on my own. I look forward to seeing what comes of this.