For an over-time analysis of 2000-2015 ACS data I would like to know why the number (and share) of married couples (both different-sex and same-sex) with at least one member having a flag on marital status (QMARST=4) increased so much in 2015. It increased from about 0.3% in 2013/2014 to 8.0% in 2015 and from 3.0% to 13.8% for different-sex and same-sex couples, respectively.
Do you know why this is the case? Was there a change in the data collection or editing procedure in 2015? What would you recommend to obtain the most reliable sample of same-sex couples?
There are two details here. First, looking into this question has uncovered an error on our end of things. Basically, there was a bug in the code that is impacting QMARST in all of the modern samples. This error will be fixed after the next data release. So, you deserve a IPUMS MUG! Email your mailing address to ipums@umn.edu to claim it.
Second, there is actually an upward trend in the allocation of the raw data of the MARST variable. This trend is independent of anything related to IPUMS harmonization, so you may want to get in touch with the Census Bureau directly about this issue.