Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: MI questions

From   Maarten buis <>
Subject   Re: st: MI questions
Date   Wed, 14 Feb 2007 08:14:23 +0000 (GMT)

--- Christopher Parker <> wrote:
> x1-x3 are not observed in dataset1, but I need them for theoretical
> purposes. Therefore, I treat them as missing data in dataset1 and
> impute them from dataset2 using demographic information common to
> datasets.
> Here's the problem: x1-x3, the imputed data, have negative values
> after merging with dataset1. There shouldn't be negative values.
> These are dichotomous 0-1 values when observed in dataset2. 

This seems to be a problem with Amelia II and not with Stata. I quick
look at the documentation of Amelia II suggests that it primarily uses
a multivariate normal to do the imputations. That is, your imputed
values are drawn from a normal distribution thus explaining the
negative values. However, all is not lost since the documentation also
states that: "Amelia includes some limited capacity to deal directly
with ordinal and nominal variables and to variables that require other
transformations". And if that doesn't work you can always use -ice-,
which is designed to deal with combinations of normal, dichotomous,
ordinal, count, whatever data. Actually, I would do both and see if
they both result in similar results, just to make sure that your
outcomes aren't the result of some artifact of the algorithm.

Hope this helps,

Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

Yahoo! Messenger - with free PC-PC calling and photo sharing.
*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index