Difference between mi append and append

[Oscar Weinzettl]

Reading through the mi append file right now and I fail to understand the difference between the two commands. Could someone please explain the difference to me.

The file says

mi append is logically equivalent to append; see [D] append. The resulting data have M = max(Mmaster, Musing), not their sum

.

What does this mean? I have 2 waves of data that I combined using append (quite new to stata and this is my first go at trying to do an empirical work) not knowing that mi append exists. In what ways have I messed up the data by using append vs mi append?

Overall I am trying to use the command mi xtset but failing horribly, so I am trying to work my way backwards and find what I have messed up. Thanks for any help!

[Clyde Schechter]

If you started out with the two data sets -append-ed together before you ran any of the -mi- commands, then what you have is all right and you don't need to do anything more or worry about it. If, however, you first did multiple imputation on the data sets and now you want to append them, you must use -mi append- so that the metadata that Stata uses to manipulate multiply imputed data sets will be correctly updated and the original and imputed variables will be correctly organized.

So it's simple. Before -mi- has touched the data, use regular -append-. If the data have already been processed for multiple imputation, use -mi append-.

[Oscar Weinzettl]

Then I have messed up, thank you Clyde!