0
$\begingroup$

In a longitudinal hospitalization survey dataset, where patients are asked to fill out a survey each time they are admitted into the hospital, one of the questions is no longer asked. This question has been answered for 10 years prior, and was removed 2 years ago, meaning that since that questions has been removed, within the longitudinal dataset there are NA's or missing data.

Would this be consider missing not at random (MNAR) since the reason for removal from the survey is not captured in the actual survey questions (AKA its reason for removal is unobserved)? This however affects every patient admitted into the hospital going forward which makes me consider if this is really MNAR.

Improve this question
$\endgroup$
2
  • $\begingroup$ This does not really answer your question, but if the question has been removed because it has been found to lacking reliability or validity, perhaps you should consider not using this variable at all in your analysis. If the reason for no longer asking the question is not explained in the documentation accompanying the data, ask the people who shared the data with you. $\endgroup$ Commented May 9 at 5:51
  • $\begingroup$ I would rather think that it could be considered missing completely at random (MCAR) since it sounds like removal of the item has nothing to do (= is uncorrelated) with the study variables. (From your description, it appears that it was not systematically removed for only some patients, e.g., those with a high or low score on some variable.) However, this may not help you since it sounds like the item is missing for everyone past the deletion date. So there is not even partial data available after it was dropped. $\endgroup$ Commented May 9 at 19:52

0

Reset to default

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.