There are lots of public data sources that report at a ZIP code level.
Of course, I can use crosswalks to relate ZIPs to ZACTAs,
but I'd really like to allocate ZIP data onto Block Groups geographies.
Is there a best practice for projecting dissimilar geographies onto one another?
(using shape files and their proportional overlaps?)
Zips are pretty big to crosswalk to block groups (62 zips in all or part in Chicago). HUD distributes zip-to-tract and tract-to-zip crosswalk files at their web site. Get the right ones for the 2010s and the 2020s.
There are many ways to adjust zcta data "down to" blockgroups. This simplest way to do this is to scale by the factor block group population/ zcta population. However you can do much better than this. You can build a model (multiway tabulation) using PUMS data for the PUMA containing the block group. You then scale (rake) the PUMA table using ACS tract and block group B S (etc) tables at the census tract and block group geographies. Usually you can locate the microdata that was used to create the ZCTA statistics/tables. If you want to know how to do this contact me at info@dorerfoundation.org. If you are working for a government or 501(c)(3) entity consults are free. Geographic "correlations" between census geographies (cross walks) can be downloaded from the Geocorr website. Dave Dorer
Thanks Dave. Very helpful