I’m using the OCC1990 variable to track how the share of workers in each occupation who have a bachelor’s degree or higher evolves over time (from 1990 to 2019). My understanding is that the OCC1990 variable is designed to be comparable over time, but it seems that there are some categorization changes between years. For instance, the case count for the “Chief Executives and Public Administrators” category jumps dramatically between 1990 and 2000, from just 255 observations in the 1990 Census to 12,339 in the 2000 Census. (The OCC2010 variable has a similar problem.) This seems like it could only be the result of a definitional change. Has anyone developed a crosswalk to reconcile these categorization changes in the OCC1990 variable across time? Or is there another way to address this issue?
You’re correct that OCC1990 is not perfectly comparable over time due to the use of different occupation codes in different years. I recommend reading the full variable description for OCC1990 and the links therein for details on how this was constructed. The simplest way to see which original OCC codes are linked to each OCC1990 code in each year is to use the method detailed in this blog post. You can then examine the definitions of the component codes in 1990 vs 2000 (for example) to get a better idea of why the number of “Chief Executives and Public Administrators” jumps so sharply.
I will also note that in that particular case, IPUMS has previously identified an issue with the original crosswalk, in that the original 1990 code (but not 2000 code) only included chief execuitves IN public adminstration, which is the main reason it jumps so much. The IPUMS USA team will be looking to update the OCC1990 coding for this issue but there is currently no timeline for this to be released.