Bonus update! Choropleths are still visually distorted and overrepresent land, so we can instead use points that are sized by population. I added an example at the end to show one way to do this with {sf}'s magical st_centroid() #rstats https://www.andrewheiss.com/blog/2025/02/19/ggplot-histogram-legend/#bonus-use-points-instead-of-choropleths
Comments
Have a look at {legendry} for a nicer circle size legend
https://teunbrand.github.io/legendry/reference/guide_circles.html?q=circle#null
Have a look at this wonderfull package
https://github.com/andrewheiss/compassionate-clam/blob/da10e854395a58cecb6c936f41394918a5d8e198/manuscript/manuscript.qmd#L483
https://stats.andrewheiss.com/compassionate-clam/manuscript/output/manuscript.html#measurement-and-descriptive-analysis
...though, wow, the plains and New England are doing way better than I expected, and the Central Valley is doing far worse than I expected.
The Mississippi and Western Appalachia are right where I thought they'd be though.