The information for the 1920, 1930, and 1940 Preliminary Complete Count Data says that “Most inconsistent information was not edited for this release, thus there are observations outside of the universe for many variables. In particular, the variables GQ, and GQTYPE have known inconsistencies and will be improved with the next release.” Whate are these inconsistencies? I am espcially interested in using information on correctional institutions.
A more detailed discussion of the comparability issues with GQ and GQTYPE can be found in the comparability tab of each variable (GQ/GQTYPE). Additionally, when extracting/analyzing the data, quality flags QGQ and QGTYPE are available for some samples; if you add those to your extract, you can see if original data was altered or not.
You can also see in the errata and revisions section that these variables have had several updates/changes. Most notably, they were improved in 2016 (see November 2, 2016).