Open JohnCRuf opened 11 months ago
Narrator: Nothing is guaranteed to work.
Copied the border design and got some really weird result: Borders are emphasized not vice versa.
Ok this was a terrible idea from the start.
Concept of a "core" vs "border" precinct is absolute garbage when half your wards look like above.
Chalk this up to a lesson in "blindly copying other people research designs in a rush without thinking them through is a terrible idea, john" take 1000
What I'll likely need to do is come up with a mapped mesh which will likely take explicit collaboration with Olivia to figure out what is going on here so I'm going to close this for now.
Ok I quickly came up with a fix for this:
Just geomatch to the census block level:
This isn't using any support data so we don't need to restrict ourselves to the precinct level. We can go deeper.
Currently working on this, should be ready early tomorrow. Getting things to the census block level took a few hours and the big problem is computing the precinct-to-nearest ward distances. The total computation will take approximately 7 hours so I'll run it when I go to bed tonight.
Olivia Bordeau has a brilliant JMP that I can copy for this project, that gives me a research design basically guaranteed to work:
https://www.oliviabordeu.com/papers/fragmented_cities_obordeu.pdf
Look at borders, order them by size. Larger wards will be less likely to spend menu money on local fringes. Use a border discontinuity design to estimate the border jump from dense to non-dense wards.