Sure. One of the big advantages of multiple imputation is that you can use it for any analysis.
It’s one of the reasons big data libraries use it–no matter how researchers are using the data, the missing data is handled the same, and handled well.
I say this with two caveats.
1. One of the steps of multiple imputation is to combine the analysis results from the multiple data sets. This is very easy for parameter estimates, but it’s a big ugly formula for standard errors. Any software that does multiple imputation should do this combination for you. So, even if it’s theoretically possible, not all software will combine the results easily for you for all analyses.
2. Censoring, which is related to missing data, but not the same, is common in survival analysis. You wouldn’t want to multiply impute the censored data that occurs naturally in the survival analysis. Survival analysis has already come up with very good solutions to censoring.
This post is part of a series of answers about missing data that I was asked during a recent webinar. There were nearly 300 people on the live webinar, so we didn’t get through all the questions. So I’m answering some of the ones we missed here.