Hi everyone,
I have several (potentially stupid) questions about multiple imputation. I've tried reading up about the multiple imputation commands but am still pretty confused...

Is there any functional difference between running mi stset, and stset alone? For instance, will there be any difference in the actual data between ‘mi stsett-ing/stset-ing’ two identical datasets; one that had multiple imputation done, and one that did not have multiple imputation done?

For some background, I have a longitudinal dataset, that is a combination of several datasets and do-files created by other researchers that I have merely merged together, and/or ran myself. Unfortunately, the final product is somehow ‘mi set’. This has created some problems when I try to stset, tsset, etc. or reshape the merged dataset. It brings up the following error:

“no; data are mi set. Use mi stset to set or query these data; mi stset has the same syntax as stset”.

Now, as far as I can tell, none of the researchers actually used multiple imputation themselves. My only theory is that, in creating their own dataset(s), one of the researchers (potentially unknowingly) merged in a different 'mi set' dataset where the mi variables had already been dropped (_mi_miss, etc.), before handing it off to me.

Does that mean that merging a ‘non-mi-set’ dataset with an ‘mi-set’ one automatically makes the product of the two 'mi-set'? I.e does an 'mi-set' dataset 'infect' every other dataset it comes into contact with?

Essentially, I’m in this weird pickle where the dataset I’m working with is 'mi set', but all of the actual mi variables were dropped by someone long before me. Therefore, I can’t mi extract, mi unregister, mi unset, or any other mi command since the variables are no longer there. I just want to be sure I’m not editing or changing the data by using mi stset, since I didn’t mi extract (or some other mi command) first.

Do I have to worry about any of that, or can I freely mi stset the data without worry of changing it?

Thanks,
David