Definitions of occupation groups in NHGIS data tables

I understand that you are looking for a list or description of the exact occupation titles included in the occupation categories used in the 1970, 1980, and 1990 NHIGS tables you listed. I also understand that you want to use comparable occupation definitions over time for tact-level analyses. There are several occupation categorization/classification schemes at play here:

  • Occupation classification schemes used to code reported occupations over time (i.e., Census Occupation Coding schemes)
  • List of occupation titles and how they are aggregated into the occupation codes (i.e., reported and used in the microdata)
  • List of census occupation codes that are aggregated into the broader occupation categories in the summary files (i.e., which microdata codes correspond to which categories in the summary files)

First, I want to make sure you understand that since NHGIS data are aggregated data, you will not be able to modify the occupations or individuals included in those tract-level counts to include or exclude certain occupations to make the population counts more comparable. Tract-level microdata (person-level) data are not available to the public after 1950, so the tract-level tables available are the full extent of what you can see at the tract-level for those years. Since the Census Bureau did not use a harmonized or comparable occupation classification scheme across those years, the aggregated data from NHGIS cannot be made comparable across multiple years, or comparable to microdata, in terms of occupation classification.

The technical documentation for the summary files (the data NHGIS provides) include information about how each occupation category is created (which occupation codes it includes). You can find these technical documents on the list of all NHGIS tabular data sources here.This documentation does not seem to be available for 1970, unfortunately, but you can view the original published Census tables from 1970 which include more information about those tables.

The Census Bureau provides an Alphabetical Index of Industries and Occupations for many census years. These indexes list all occupation titles that are coded into the broader occupation codes in the census microdata. See this IPUMS Forum post for links to the indexes from 1940 to 2016. It is not clear to me how the occupation categories are constructed in the aggregated (NHGIS) data. They may be constructed from occupation titles or from the microdata occupation codes. They may also incorporate information on industry or class of worker.

While you can’t change which occupations are included in the summary data occupation groupings, you have a couple options for how to code occupation in the 1950 microdata you are using. I suggest taking a look at the occupation classification scheme for the 1950 census as well as the classification scheme for IPUMS-constructed variable OCC1990, which reports the occupation of each individual in terms of the Census Bureau’s 1990 occupation classification scheme. OCC1990 is available from 1950-onward. The 1990 classifications may be more comparable than the 1950 classifications to those of the 1970 and 1980 censuses. If so, you could use OCC1990 with the 1950 microdata to make some gains in comparability between 1950 and later years. That is, you can make the 1950 microdata comparable to one or more years of NHGIS data. Reviewing the occupation classification schemes for each decade may also provide some insight into how much certain occupation categories included in the NHGIS tables have or have not changed over time, and how much of a comparability issue you really have.

Census tracts are population-based geographic areas that include about 4,000 residents each. Their definitions change, usually every 10 years, to accommodate changes to the size and distribution of the population. This means that census tract boundaries change over time. IPUMS NHGIS offers crosswalks between different definitions of census tracts, starting with 1990 tracts. We also provide GIS files of census tracts from 1910-onward. To find the GIS files, in the NHGIS data finder, filter only on geographic area (census tract) and then click the GIS Files tab. I would highly recommend reading through both of these past IPUMS Forum posts where users have asked relevant questions about using tract-level data from before 1990: Longitudinal data crosswalking follow-up and Joining longitudinal census data with spatial data.