Coverage of counties in 1950 census onwards

I have been working with the 1930 and 1940 censuses, in Stata format, at the individual level. Then, I have collapsed the data to the county level easily and have obtained what seems to be all counties at the time.

I want to extend the analysis to the 1950-1970 censuses, but these ones seem to be curtailed. After collapsing the individual-level data to county level, I only get 145 counties (the same exercise in 1940 yields 3,065 counties).

Am I selecting the wrong file, or are the 1950+ individual level censuses limited on the number of counties represented?

P.S: If anyone wonders why I use individual-level census if my unit of analysis are counties, the answer is that the variable I need (share of the labor force engaged in agriculture) can be constructed only from the individual level, as far as I am aware.

Thanks

Counties are not identified in public-use microdata from 1950 onwards, so IPUMS instead identifies counties, where possible, from other low-level geographic identifiers. These include State Economic Areas (SEA) in 1950; county groups in 1970 (CNTYGP97) and 1980 (CNTYGP98); and Public Use Microdata Areas (PUMA) from 1990 onwards, including Super-PUMAs (PUMASUPR) in 2000. In 1950 and later samples, COUNTYFIP identifies a county if and only if:

  1. it was coterminous with a single SEA, county group, or PUMA; or
  2. it contained multiple SEAs, county groups, or PUMAs, none of which extended into other counties.

You can read more about county identification and view a list of identified counties by sample in the COUNTYFIP variable description. Note that all counties are identified in the full count 1950 census data, available publicly from IPUMS USA. If the 1950 full count data are appropriate for your purposes, then you could consider using them in order to obtain information on all counties. See the notes on the 1950 full count data, including known issues, on this page: IPUMS Full Count Data (1790-1950).

IPUMS NHGIS provides summary data from the U.S. decennial census, American Community Survey (ACS) and other sources, aggregated at a variety of geographic levels, including county. From IPUMS NHGIS you can obtain county-level data that report the number of persons working in agriculture in select years, including 1820, 1840, 1980, and 1990. In other years, agriculture as an industry of work is often combined with other industries in these summary data, making it impossible to distinguish between agriculture and, e.g., forestry or fishing. Industry categories are typically more detailed in the microdata. I am assuming that your definition of “being engaged in agriculture” refers to an industry, but note that working on a farm encompasses a smaller range of occupations. In some additional years, data are available via IPUMS NHGIS that identify sets of agricultural occupations (e.g., farmers and farm foremen), which may or may not be acceptable for your purposes.

Depending on the scope and scale of your project, you may be interested in the 1960 published census volumes, which tabulate occupation and industry by county, where agriculture as an industry is separately identified. These are scans of original Census Bureau published summary data, and have not been transcribed by IPUMS. Using this resource would be time-consuming.