Re: st: Multiple imputation with panel data

[weddings@stata.com (Wes Eddings, StataCorp LP)]

Veronica Galassi (V.Galassi@sussex.ac.uk) asked about performing multiple
imputation on panel data:

> I am trying to impute my dataset, an unbalanced panel with observations over
> two years (2003 and 2007), using multiple imputation techniques.
>
> (snip)
>
> In particular, after reshaping my panel from long to wide form, I tried to
> impute it using the ICE user-written programme which I have read performing
> better with panel than -mi- command.  However, I got back errors like the
> r(2001)saying "only 1 distinct value of year_d2003".
>
> (snip)
>
> I have also tried to impute the dataset through the multiple imputation
> control panel choosing the -mi impute mvn- command.  In this case, what I got
> back was the error message r(2000)saying "No observations to obtain initial
> values for EM using available-cases (AC) method." 
>
> I do not understand what I am doing wrong because I told Stata that I am
> working with panel data and I set the panel id variable... 
> 
> (snip)

Neither -ice- nor -mi impute- has an imputation method specifically designed for
panel data.  (The -mi xtset- command does declare panel data but does not change
which imputation methods are available.)  We do, however, have a FAQ that has a
few suggestions for applying -mi impute- to panel data.  The link is

http://www.stata.com/support/faqs/statistics/clustering-and-mi-impute/

Veronica also mentioned that -mi impute mvn- returned the error r(2000), "No
observations to obtain initial values for EM."  Veronica may consider providing
her own initial values to the EM algorithm, using the option -initmcmc(em,
init())-.

--Wes
weddings@stata.com


*
*   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/