Spouse hours of work and marital status

I’m looking at marital status and hours of work of the spouse for ASEC samples and find some inconsistencies. For example, variable uhrsworkly_sp (hours of work of the spouse last year) pernum 2, serial 41938 in years 2007 and 2008. (I understand that they are different individuals, I’m just using the same serial and pernum for the example below.)

  1. In 2007, uhrsworkly_sp has a value of 999, and marst (of the reference person) is 1 (married). That looks fine, it could be the case that the spouse never worked last year.
  2. But in 2008, uhrsworkly_sp has a value of 50, yet marst is 6 (never married/single). If the person was never married, how is it possible that uhrsworkly_sp is 50?

Situations like 2. above show up in 2.77% of my extract cps_00024. How should I interpret that? does the label “never married/single” mean “never married” or “single”, with “single” being interpreted as currently single but could have been married before? or is that an error in the data? I noticed that those inconsistencies only show up in my data starting from 1995. Was there a coding change then?

I agree with your assessment of the 2007 case where the person is married but their spouse has a uhrsworkly value of 999; the spouse may not have worked last year. The second situation is unmarried partners. Their marital status (appropriately) reflects that they are not married (and specifically have never been married), but IPUMS assigns them as a spouse based on their relate value. I am including a truncated tab of UHRSWORKLY and RELATE for persons who have a marital status of 6 but do have a spouse assigned via SPLOC (which is required for the attach characteristics feature to work) for all ASEC samples.