[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: multiple_imputation

From   Maarten buis <>
Subject   Re: st: multiple_imputation
Date   Sat, 11 Jul 2009 09:13:54 +0000 (GMT)

--- On Sat, 11/7/09, Victor Mauricio Herrera wrote:
> I am working with a pool of surveys and I want to impute
> missing values in the pooled dataset while keeping the
> design variables and re-calculated weights. Here is my
> question: Is it correct to impute missing values - and used
> them during the analysis - for variables that were not
> originally measured in some of the surveys? For example,
> waist circumference was not measured in all the surveys,
> however, since that variable must be included in the
> imputation model all individuals in the pooled dataset end
> up with a value for that variable. 

In general I like this approach; this way you can also use 
the information available in surveys that only asked some
of the questions. In practice getting a good imputation 
model can take quite a bit of puzzling. In particular 
trying to incorporate characteristics of the surveys can
become quite tricky. If you need a reference of someone
who has done this before see (Gelman, King, and Liu 1998),
though their exact imputation model has not been implemented
in Stata.

Hope this helps,

Andrew Gelman; Gary King; Chuanhai Liu (1998) Not Asked and 
Not Answered: Multiple Imputation for Multiple Surveys. 
Journal of the American Statistical Association, 

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen


*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index