Re: st: Missing Values

[Fred Wolfe <fwolfe@arthritis-research.org>]

At 05:10 PM 7/7/2003 -0700, you wrote:
> Does Stata have any way to impute missing values using expectation
> maximization (EM), or a multiple imputation approach?
> Alan Acock
findit impute identifies the following multiple imputation methods:

STB-54  sg116.1  . . . . . . . . . . . . . . . .  Update to hotdeck imputation
        (help hotdeck if installed)  . . . . . . . .  A. Mander and D. Clayton
        3/00    p.26; STB Reprints Vol 9, p.199
        options added for allowing input of datasets imputed by
        other programs and controlling the random number seed

This is a multiple imputation program that works nicely in Stata

STB-45  sg92 . . . Logistic regression for data including multiple imputations
        (help implogit if installed) . . . . . . . . . . . . . . . . . C. Paul
        9/98    pp.28--30; STB Reprints Vol 8, pp.180--183
        program that automates the combination and correction of
        coefficients and standard errors for multiply imputed data
        in the logistic regression context

There is also a program by Gary King ("amelia") that does EM. Caution it 
can run for days! on large data sets. I turned it off once after 36 hours.



Fred Wolfe
National Data Bank for Rheumatic Diseases
Wichita, Kansas
Tel (316) 263-2125     Fax (316) 263-0761
fwolfe@arthritis-research.org


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/