Hacker News Reader

Comment by abeppu | Hacker News Reader

Did I miss it or is there no methodology description justifying how they reached this?

If there was some good standard survey on cultural views, you could compare geo regions on the summary stats of their responses, and cluster them. But you'd need a _huge_ number of responses to get good county-level data. And then I think we'd expect to see lots of county-to-county differences reflecting the urban-rural contours, immigration differences tied to industry, etc, rather than these big, uninterrupted regions. E.g. I would think King County, WA and Alameda County, CA have a lot more in common with each other than either does with Del Norte County, CA.

View on HN

You didn’t miss it, because there was no methodology to speak of. Colin Woodard isn’t a sociologist or historian, he is a reporter for the Maine Sunday Telegram who wrote a book premised on the idea that the arc of 300-400 years of culture and history of the various regions of the US are cleanly defined entirely by the initial groups that settled them. Every region has a stupid, manufactured brand (it’s not the West Coast and New England, it’s the “Left Coast” and “Yankeedom”). What little data that is referenced is cherry picked to support the narratives. It’s also hilariously patronizing (spoiler: The reporter from Maine concludes that New England and the West Coast are the epitome of all that is clean and good, and everyone else are barbarians and unwashed rubes).

It is closer to a Buzzfeed quiz explaining how your astrological sign dictates your Hogwarts House than anything remotely resembling academic rigor.

Thank God, so I'm not going crazy. Calling the Northeast (Irish-Italian) and Great Lakes (German) Megaregions a single country called "Yankeedom" because they both lean Left made absolutely zero sense to me.

i'd like to see more methods, because that Midlands region looks similar to the types of artifacts I used to get doing various unsupervised learning projects.

that shape screams "there are a couple of clear clusters nearby, and this is the leftover 'in-between' space we didn't know how to handle so we made a new cluster"

the arbitrary New France/Deep South selections near New Orleans and Baton Rouge are also quite amusing

I thought this too, it seems to be a map of who the original people were to the region (including cases where major immigrant groups who created the region), but none of the methodology is listed or really makes sense, making it not very useful. Seems rich of the author though to suggest the region he's from is deeply complex, while the regions he isn't from are painted with a broad brush

No, it's not in the text at the original link, however, yes, there is a link to the full article in the blurb with the how: https://www.nationhoodlab.org/the-american-nations-regions-a...

No, I saw that page and it doesn't say anything meaningful about how they arrived at this. It says _that_ they expanded the model, and various notes about resolution, they use to communicate the labeling they arrived at -- but nothing meaningful about the input data used, any statistical methodology, etc.

> my Motivf colleagues and I refined the ad hoc models and produced what you might call the “official” American Nations Model spreadsheets for the United States, mapping the regional cultures at county-level resolution.

> This summer, we’ve expanded the analytical model to the rest of North America covered in American Nations.

There's an entire book from the author on this topic [1]

[1] https://www.amazon.com/American-Nations-History-Regional-Cul...

I've read the book and it's interesting. I don't recall him explicitly explaining the methodology and so I had the same question - from that link it sounds like the methodology was more ad hoc originally but now it's on more of a solid quantitative footing.

Reply on news.ycombinator.com