Where are New England and the Midwest?
Who lives in The South? Is Pittsburgh on the East Coast? These are the questions I occasionally find myself asking, any time I see or hear someone name a region but with a clearly different definition in mind than the last person who referred to it. More so whenever I encountered people actively arguing about whose definition was “right” or “true”. Eventually, despite my complete lack of qualifications as a professional statistician or survey-writer, I decided to gather some data. I took the map drawing code used by Bostonography for their 2013 neighborhood survey and adapted it for my needs, then asked a bunch of people to fill it out. That was eight years ago, and rather than let the project sit forgotten I figure it’s time to actually write up some of the results.
I am intentionally not linking to the project here. Although you can surely find it if you search, I want to avoid sending people to it after seeing the results here has spoiled them, and also I am hopeful to relaunch a much better version of the project if this post inspires a data scientist or UI/UX person to volunteer help improve the science and/or usability here. Sadly a lot of the images below don’t very effectively differentiate between 50% and 90% of people selecting an area, so please let me know if you want to help fix that.
I advertised my survey in my personal circles, on Facebook, Google+, LiveJournal, etc. I also posted it to a few mapping-related forums, including on reddit. A few people tweeted about it, there was some discussion in a few niche parts of the internet, and even a few small unsolicited articles and blog posts. Definitely not the most bias-free way to distribute a survey, but it got the word out. There are likely strong biases toward educated people, people of liberal politics, people who live(d) in Nashville, Atlanta, or Boston, people who frequent GIS and mapping forums, and people who attend fandom conventions. The instructions and UX were also very amateur and not thoroughly tested or optimized, so the data likely contains errors related to that as well.
All that said, nearly a thousand people did submit some regional boundary drawings, with some clearly discernible patterns.
Despite my lack of skill with rendering, you can see here that most outlines were drawn along state lines. Some notable exceptions are people cutting off the west tip of Texas, sweeping diagonals across the Midwest, and the almost complete lack of distinction between Washington and Oregon, or Vermont and New Hampshire.
Backing up a moment to the question of bias, one of the “regions” on the survey was named “Home”, and users were given the instruction to draw the place they grew up, where their understanding of regions and their names would have developed.
Some people clearly drew whole states or regions here, while others narrowed down to pretty small circles around cities. Despite lacking all the necessary tools to query this data, I can see that I got dozens of people from each of CA and TX and the midwest, so perhaps my sampling bias wasn’t quite as bad as I had feared.
To start things off, let’s talk about compass directions. These are regional labels like North and Southeast. If I asked you to outline the West region of Brazil or the Czech Republic, you could probably take a crack at it even if you’ve never seen those countries on a map. So, let’s see where people drew those lines in the US.
Where are the East and West parts of the country?
I opened with this pair because the division is very clean. The East smoothly declines from 95% in Pennsylvania in every direction, to 85% in Maine, 65% in Florida, 15% in Louisiana, and 10% in Wisconsin. The West is even more predictable with an almost perfect eastward gradient along columns of states. The west coast states are 90%, holding strong to 75% for the line from Minnesota down to New Mexico, then a sharp drop to 20-30% for the next column of states from the Dakotas down to Texas, and a final drop to 5% for the column from Minnesota down to Louisiana.
How about the North and the South?
More so than East and West, North and South have historic meaning and political connotations, as well as association to personal identity and heritage. From the graphic you can see there’s a lot of contention over Missouri, Nebraska, Maryland, and West Virginia, and less for Kentucky and Virginia. A less obvious trend hinted by the bright white line north and west of Texas and Kansas is that there’s also an east/west component, which probably has something to do with which states existed around the time these distinctions were being cemented in our national zeitgeist. While 70-80% of people included Maine to Wisconsin in the North, there’s a sharp drop to 60% in the Dakotas with a gradual decline to 30% including Washington. The South is even more sharply defined, with 90%+ including Virginia to Louisiana (except for Florida which only fits for 60-80% of respondents), 60% for Texas, then 20% in New Mexico dropping to 5-8% in southern California, which was also labeled as the North by 2-4% of people in ways that are visibly clearly intentional!
The East and West Coasts are relatively clear.
There’s a roughly even split between people who drew some distance inland from the coastline vs people who followed state boundaries. Among the state boundary folks, we have 15% of people labeling Pittsburgh East Coast, and 25% putting Spokane on the West Coast. The 5-10% who intentionally included Nevada on the West and/or West Virginia on the East are notable outliers.
How about the diagonal compass directions?
When asked about the Northeast, Southeast, Southwest, or Northwest, the expected results are there along with some that I found surprising. Firstly, unlike the orthogonal compass directions, this map has a hole, with most of the Midwest (see below) not being represented. Despite having some significant representation in the East, West, and North categories, Wisconsin and its neighbors all the way down to Missouri are mostly left out here. Somewhere around Sacramento is where California switches between majority Northwest vs Southwest, while all the other tipping points very sharply follow state lines.
With those relatively straightforward examples out of the way, I want to cover the two regions that actually inspired this project in the first place. These are regions for which I’ve recognized hundreds of discrepancies across thousands of times I’ve heard them used, and I’ve seen dozens of arguments about, most of which I wasn’t party to. Without further ado...
The Midwest
With a peak of just 85% in western Illinois, this map gives form to what I’ve recognized my whole life. People mean different things when they refer to the Midwest, including some entirely disjoint meanings. Although I can’t render it, this is one of the only regions where I tried to manually dive into the correlation with where people drew their Home. My amateur conclusion is that the definition of the Midwest is “halfway between where you grew up and the West”. Almost no one from New England included Texas (5-10%) in this region. People from the South tended to not include the Great Lakes area (60-70%). And the people who included states as far west as Idaho (5%) and New Mexico (5%) tend to have grown up farther inland than the East Coast.
New England
First a reminder that this rendering is not proportionally color accurate. The outlines are correct, and the gradient does map monotonically to percentage, but not linearly.
This is the region for which there is the greatest agreement on definition, and the second most vehement opposition to disagreement (the South takes the cake, there). Most people know it to be exactly six states, and refuse to even humor the possibility that anyone who thinks otherwise can be anything but wrong or mistaken.
98% of respondents included western Massachusetts in their drawing of New England. I take this as a useful noise floor for the whole survey, and will assume it has at least 2% error (likely more, of course). Vermont and New Hampshire (which are effectively the same place as far as this survey is concerned) keep it above 90%, with a drop to 80% in Maine. Maine actually goes significantly lower toward the tip, but I can’t tell how much of that is intentional vs imprecise drawing. Connecticut has a few lines across it that look intentional, dipping down to 90%. Rhode Island is too small and close to the edge, it's overwhelmed by the noise of imprecise coastal lines. That covers the “official” six states, which 80-100% of respondents agree with.
That leaves the rest of the pink on the map...
28% of people included northern New York in the New England region, and 19% for the western parts of the state. Sadly the drawings don’t have nearly enough precision to see how people feel about NYC. New Jersey was included by 20% of people, and Pennsylvania by 8-16%. Maryland and Delaware are too small and hard to follow, but they are generally around 10%. And finally, 2-5% included the Virginias, including at least ten people who very clearly drew along the southern border of Virginia.
If I get good feedback on this post, and/or if someone steps up to help work on better data analysis or interactive tooling, I’ll make a followup post that dives into some more specific compass-related regions like the Pacific Northwest and Deep South, as well as a wide variety of other regions like Appalachia, the Rockies, the Bible Belt, the Great Lakes, and the Gulf Coast. However, for the moment I will stop here.
Maybe eventually this data and article will be considered reputable enough to get into a certain online store of knowledge that currently declines to acknowledge that there is any disagreement on the definition of this region. Hint, hint, wink, wink, nudge, nudge.
I am posting this in all the places that I post blog-y stuff, but I don't really have an outlet for anything approaching an article. If you think there's somewhere else I should post or submit this, let me know.