Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# AW: st: reporting results from a multiple imputed data set

 From "Lamla, Bettina" <[email protected]> To "[email protected]" <[email protected]> Subject AW: st: reporting results from a multiple imputed data set Date Wed, 20 Apr 2011 15:52:16 +0200

```Hi Joerg!

Well, regarding

(1) i can only include  M-1 year dummies.
If my command includes M year dummies anyway, stata will omit one dummy due to collinearity, i.e. the least significant one.
(2) the least significant one varies across imputations. Hence I could just exclude let's say year 1 for all estimations and hence cannot report an estimate for this year
OR
I let stata pick the year dummy to be excluded and as a result -after applying the rubin rule- have estimates for all M year dummies even though this would not be possible in the individual regressions

Hope this clarifies

-----Ursprüngliche Nachricht-----
Von: [email protected] [mailto:[email protected]] Im Auftrag von Joerg Luedicke
Gesendet: Wednesday, April 20, 2011 3:43 PM
An: [email protected]
Betreff: Re: st: reporting results from a multiple imputed data set

On Wed, Apr 20, 2011 at 9:24 AM, Lamla, Bettina
<[email protected]> wrote:
> Hi everyone,
>
> I need a recommendation.
>
> I have a multiple imputed data set (N imputations) and with a set of dummy variables (M years).  Let's take year dummies as an example. Stata omits of the M year dummies due to collinearity.
>  As it turns out it is not always the same year dummy in the N regressions.  In conclusion, when I apply the Rubin Rule I get an estimate for all N year dummies.
>
> Which way is the correct one: Shall I just exclude one dummy prior to running the regressions or do I report (after application of the Rubin Rule) coefficients for all M dummies?
>

I do not understand several things here:

1) You say "Stata omits of the M year dummies due to collinearity".
What exactly is getting omitted?

2) "As it turns out it is not always the same year dummy in the N regressions".
So for any given subject, the dummy should be constant across all N
(btw usually it is referred to the number of imputated data sets with
the letter "M"). What exactly do you mean with "not always the same
year dummy"?

3) "Which way is the correct one: Shall I just exclude one dummy prior
to running the regressions or do I report (after application of the
Rubin Rule) coefficients for all M dummies?"
Are all dummies indicators for measurement (or whatever) years? So why
do you want to exclude any year indicators? I feel this might be
related to 2).

J.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

Rheinisch-Westfälisches Institut für Wirtschaftsforschung e.V. (RWI)
Hohenzollernstr. 1-3
D-45128 Essen

Phone: +49-201-8149-0
Fax: +49-201-8149-200
Web: www.rwi-essen.de

Executive Board/Vorstand: Prof. Dr. Christoph M. Schmidt, (Präsident),
Prof. Dr. Thomas K. Bauer (Vizepräsident), Prof. Dr. Wim Kösters
Registration-No./VR 1784 beim Amtsgericht Essen

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```