Re: st: Multiple Imputation (MI)

[Alan Acock <acock@mac.com>]

I think of secondary data as data others have collected and you are using. The choice of using multiple imputation has to be weighed against the alternatives. The default alternative that most procedures apply is casewise (listwise) deletion. This default assumes missing values are missing completely at random where multiple imputation only assumes missing at random. While multiple imputation may not be justified, it will be more reasonable than the default approach. Of course the more you know about the data the more options you have and if I have the same understanding as you do about what secondary data is, then you may have limited information about the data.
Alan Acock
acock@mac.com



On Jan 13, 2012, at 8:28 AM, john ebireri wrote:

> Dear Statalist Users,
> 
> I just want to ask if MI is suitable only for primary data
> 
> I have read a few papers on MI and they keep talking about it used for missings in survey data.
> 
> Is it suitable for this method (MI) to be used in handling missing data in secondary data?
> 
> Thanks.
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/