I’m having trouble figuring out what the most fine-grained level of geography is that I can extract from the 100% sample U.S. census data for 1910 or 1920. I can see county - is that really the most detailed one can get? I see there is a variable for the size of a place (SIZEPL) that seems to differentiate between several places in a county, but I can’t see any way of identifying the place itself except for the largest cities?
Thanks!!
Depending on the geographic area you are looking for, COUNTY, CITY, or METAREA may be the smallest unit for 1910 and 1920 100% samples. For the 1910 and 1920 1% samples, the WARD variable is available, which identifies the political ward where the household was enumerated. Unfortunately, public use microdata are not available at the town/municipality level.
If you are able to use aggregate data for your analysis, I would recommend that you look into the tables available through IPUMS NHGIS. IPUMS NHGIS provides spatially aggregated census data and has many summary tables available at smaller geographic levels such as census tract and zip code.