Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Limit on imputing miss values?


From   Richard Goldstein <[email protected]>
To   [email protected]
Subject   Re: st: Limit on imputing miss values?
Date   Wed, 14 May 2003 13:48:59 -0400

Using a dummy variable is definitely a problem; see
Jones, MP (1996), "Indicator and stratification
methods for missing explanatory variables in multiple
linear regression," _JASA_, 91: 222-230

Rich

rollanders wrote:
> 
> One alternativ is of course to assign the missing values a value
> which allows them to be included in the analyses as a dummy variable.
> It may be of interest to see how the other predictors perform if you
> include the dummyvariables for missing values compared to if you
> impute them. It would also be of interest to know if these persons
> are different from the others in any way. What persons are less
> willing to tell about their income? In what way may the exclusion of
> them give biased results?
> 
> Roland Andersson
> 
> --- In [email protected], "Sabine N. Merz" <uribazo@r...>
> wrote:
> > Dear All:
> >
> > I have a question where I simply do not know if there is a "right"
> answer
> > but I would much appreciate some feedback because there might be
> a "golden
> > rule" that I simply have overlooked. (Too much time spend dealing
> with this
> > would explain it.)
> >
> > *Is there a limit on how how many missing values one can or rather
> should
> > impute?
> >
> > In my particular case the missing values all concern the oh, so
> sensitive
> > personal income question. I already cross checked, e.g., with labor
> force
> > status, hours worked, and such but I am still left with about 7% of
> my
> > respondents who did not answer this question. I would not feel
> uncomfortable
> > imputing 1% of missing cases but 7% seems a bit much. From the
> missing
> > values options provided it does seem that these folks simply
> refused to
> > answer the question.
> >
> > I will likely simply use the stats "impute" option and see what I
> get and
> > compare these results to what would happen if I simply leave out 7%
> of my
> > sample. (Or would you suggest another method to deal with missing
> values?)
> > Still neither one is a very elegant solution.
> >
> > Thank you for any advice you might have.
> >
> > Best wishes,
> > Sabine Merz
> >
> >
> > *
> > *   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/
> 
> *
> *   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/
*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index