should you include DV in MCAR when missings are found in IVs?

Question

can anyone please help me with this question, can't find a satisfactory answer on the stats sites. A reviewer of one of my articles commented that I should include the DVs in my MCAR analysis. But i only have missings in my IVs. thanks

Answer 1

There is some guidance in this article by Sterne and colleagues in the BMJ. They point to a cautionary tale where not including the outcome in the imputation model was potentially mis-leading. So the reviewer has a point.