Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
-------------------------------------------------------------------------------- Well, as the output shows there is only one observation (and this is what you ask for). You cannot estimate a standard error from one observation. To be more precise, since the within-imputation variance cannot be calculated (because there is only one observation) and therefore returns missing, the total variance is also missing. This follows from the simple fact that T = W + (1+1/M)B, where T is the total variance, W is the mean of the within imputation variance, M is the number of imputations and B is the between-imputation variance. If you (for whatever reasons) are interested in the between-imputation variance I suggest merely using -summarize- or the like, as -mi- will probably not help here. Best Daniel -- I have multiply-imputed data: the "SWIID" data, downloaded from http://myweb.uiowa.edu/fsolt/swiid/swiid.html. The data consist of country-year observations (173 countries, annually 1960 onwards). For each observation, there are 100 multiply-imputed estimates of "gini_net" (Gini coefficient of income inequality). [...] I would like to derive MI estimates of gini_net for a particular year, say Denmark in 1989. [...] The "Std. Err." estimate is missing -- why? Please access the attached hyperlink for an important electronic communications disclaimer: http://lse.ac.uk/emailDisclaimer * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/