Restriction in imputed results (Multiple Imputation regression)

[Ishmam Al-Quddus]

Can anyone help me, how to restrict the imputed results from multiple imputation regression greater than 0 in STATA.
I have missing data in the variable 'distance', hence, I need to have the predicted value greater than 0. I got few negative values after the imputations.

[daniel klein]

Usually, pmm works quite well; make sure to request more than one neighbor to be matched.

Best
Daniel

[Ishmam Al-Quddus]

Hey Daniel

I have tried pmm as per your advice. After imputations, I have seen the number of Data in my data set increases 6-7 times. I started with 400 data. In two variables, there were 54 and 57 missing data. After imputation, whenever I browse the data set, I see around 4800 data and there are still missing data in my data set (represented by "."), which I thought would be replaced by values generated by pmm.

[daniel klein]

The "problem" you describe in #3 has nothing to do with the imputation method pmm; you have observed this before.

Please do read the pdf manual on MI to learn all about MI in Stata, including different styles in which imputed datasets are stored. Perhaps start with

help mi styles

but then go on to the full pdf documentation. Do not worry about to run in a series of technical details, the pdf is really easy to read with lost of examples (of course, the technical details are there if you want them).

Best
Daniel