Hello!
I’m writing to ask about the partnership history of NEPS 6. I tried to construct the partnership history of respondents from spPartner. However, I’ve seen a lot of negative values (-54, -55, in particular) for the variables ts3152y and ts3153y. Shall I take this as the respondents were still in the partnership? I’m not sure how to understand „not determinable“ or „missing by design“ in this particular context. Thank you very much!
Best regards,
Chen
Dear Chen,
asking people about their partnership biographies is not an easy task and therefore analyzing this data is not trivial either.
However, the distinction between these two missing values is quite clear: Whenever a certain question was NOT part of the survey, i.e. a person was not asked this question at all, you will find the value -54 „missing by design“. For example, from wave 12, i.e. in the last two published survey waves, people were no longer asked whether people had moved back together with their partner or moved away. Therefore, all people receive a value of -54 for these variables in the last two survey waves.
The value -55 „not determinable“ is assigned as part of episode harmonization (see the DataManual chapter 4.4.3) if no other valid value was found in the subspells. For the specified variables, this only occurs in observations with subspell ==0 and is actually an artifact without any substantive information. You could also replace these values by sysmis.
Of course I don’t know what your analysis goal is and how you operationalize the partnership status, but I wouldn’t use the information from ts3131y, th32363 and ts3152y to construct a partnership status because they actually only represent information about where someone lives, but not whether a partnership continues. I would rather rely on the information from th2811* (start of partnership), ts3141* (date of marriage), ts3153* (date of divorce), ts3151* (date of death partner) and th2803* (end date of partnership).
Best of luck,
Benno
Thank you Benno!!