Standard deviations with multiple imputation

[LS Mofid]

Hello all,

I am a student who is analyzing data from a randomized controlled trial. We had some lost to follow-up (<5%) and therefore I have used multiple imputation to impute the missing outcome data to satisfy the intention-to-treat principle for my primary analysis.

My supervisor would like me to provide the mean and standard deviation of the outcome measure (infant weight) for each group. Problem is, mi estimate: mean will only give me the mean and standard error. Is there a reason for this?

Any information as to why SDs cannot be calculated for MI data would be extremely useful.


Thanks,

L

[daniel klein]

One reason might be that fact that multiple imputation is meant for inference not descriptive statistics. I have written a command, misum, a couple of years ago that pools results from summarize in mi-datasets. You will have to decide whether these pooled results are meaningful in a way.

Note that the SD can be calculated after mean simply by multiplying the standard error by the square root of the number of observations.

Best
Daniel