On pre-attached spousal data

I’ve been pooling years 2003-2015 of the ASEC to look at married couples. Looking through some descriptive tabs, I noticed that some respondents who are not in the married category (e.g., divorced, widowed, never married/single) somehow have spousal data. From my understanding, spousal data was only about a married respondent’s spouse, not a respondent’s unmarried/cohabitating partner. Am I misunderstanding something? Who are these extra “spouses”?

The answer to this question may be found in the SPRULE variable, which addresses some ambiguity when forming spousal links. In most cases, a person must be “married, spouse present” according to MARST in order to receive a non-zero value in SPLOC. One exception to this rule, which you may be observing in your data, is for links between a householder and an unmarried partner. The SPRULE variable is a two digit variable that aims to clarify two types of ambiguity. Links are made within a household based on how clear the relationship is between two people based on the RELATE variable. This first digit identifies links based on their relationship to a reference person, however this link may not be unique within the household. The second digit in SPRULE provides information on the level of clarity of the spousal link. A second digit equal to 1 indicates that the pairing is very clear and a 6 indicates that the paring is not very clear.