details about errors with CPS union data from 1983-1989?


The documentation for the UNION variable notes:

“Information about union membership was collected in the ASEC CPS beginning in 1983, as part of a series of questions about usual weekly and hourly earnings. The Census Bureau reports that the results in the ASEC CPS public use files for this data series included errors for years prior to 1990, so for the March ASEC supplements only data from 1990 forward are part of IPUMS-CPS.”

Is there a detailed summary of the errors associated with the UNION variable in the ASEC CPS public use files available from the Cenesu Bureau to which you could point me? I did not see a specific reference or link to a report on these errors in the documentation, and an online search did not reveal anything that I could definitively link to the errors mentioned in this note–there has been some work on the problem of roughly 2-3 percent of CPS respondents misreporting their union status, but this strikes me as a separate and more general problem that does not uniquely affect the period from 1983-1989 and thus is not the reason motivating their exlcusion from IPUMS. If there is no such specific report, would you be able to provide a more detailed description of the errors identifed by the Census Bureau and the reasons that IPUMS judged them to be sufficiently serious to warrant excluding the union data altogether?

Thank you for whatever information you are able to provide.

I have also been unsuccessful digging up any information or report about this. You are not the first person to ask about this, so the IPUMS CPS team will be looking into this in the future.

I came to the support pages to find out the answer to this question as well! Have you found anything since 2017? I would really love to be able to use this earlier union data for my research, and I know other research uses CPS data on union status from these years, so I’m very curious to know what these errors are and how much they affect the use of these variables. In addition, if one were to try to use this earlier data, are there resources on how one would use the unharmonized versions?

No new information on these variables has emerged. You can however use the UNION data from the March Basic CPS for 1984-1989, for those who are linkable between March Basic and ASEC files.

Ok thank you! And 1983 is the earliest available data on union membership for IPUMS CPS, correct?

Yes, but only for ASEC. For basic monthly samples it is available from 1984.