NHGIS provides great shapefiles since 1790, which show county boundaries nominally, as they were in the time of each Census. But I cannot find data or code that makes such boundaries consistent over time.
I am searching for code that maps, for example, counties in 1870 onward to their parent counties in 1860. Or counties in 1920 onward to their parent counties in 1910. So then I could use units which are consistent (fixed boundaries) over time.
Is there a source for this?
NHGIS does not provide any files or code linking county definitions across time, nor do any other IPUMS projects. I also don’t know of another source for this specific type of information in a simple format.
A data set that may be helpful is Historical Population Estimates for 2010 Counties, which I generated from NHGIS data. That source supplies only total population counts, however, so if you need to standardize other types of data, you would still need some type of crosswalk. That page also supplies some documentation on the methodology used, which you could follow if you wish to construct your own crosswalks.
A U of M research project also recently constructed a crosswalk for 1970-2010 counties. I asked one of the researchers (Jasmine Trang Ha) about it, and she says she’s not sure where it is at this time, and it’s in Stata format, though she could convert it to CSV. If that would be useful for you, please let me know (replying to this answer), and I can relay the request to her.
Thanks for the reply!
I would be happy to see the U. of M. files, if possible. Particularly if there is code as well, so I can migrate them to a broader context.
If/when I generate this relative crosswalk code, I would be happy to share it with the IPUMS community.
Jasmine Trang Ha has helpfully organized her county crosswalk files for sharing, and I’ve put them in a zip file at:
A few notes:
* There are four crosswalk files that go from 1970, 1980, 2000, and 2010 to _1990 counties_.
* These crosswalk files are stored in Stata .dta format, and Jasmine has provided related Stata code and a readme file. If you’re unable to use these, please reply, and we can transform the data to CSV or another simple format.
* If you have questions about how to use the four crosswalk files, I suggest you email Jasmine directly (address available through link above).
* Jasmine, along with Jack DeWaard, intends to publish a working paper about this product soon, hopefully by the end of the summer. If you use the data in another product, you should be able to cite that paper.
* The archive also includes a CSV version of the original “complete” crosswalk I provided to Jasmine, which supplies estimated populations for every spatial intersection among the 1790-2000 definitions of counties. You could use this to generate a crosswalk between the counties of any pair of years from 1790 through 2000. (The main reason the crosswalk does not include 2010 is that it is based on the 2000-TIGER version of NHGIS county definitions, which are not consistent with the 2010 county definitions. There is not currently a version of NHGIS county definitions that covers the entire period 1790-2010.)