Hadi's distance

[Bor Kemkes]

Dear Statalist members,

I want to econometrically justify why I dropped outliers. I want to check for this possibility in the a regression where I use a measure for social expenditure as the dependent variable and a measure for openness as an independent variable (and some control variables).
In this case I want to use Hadi’s distance to see if there are any outliers in my data. I am not entirely sure what code I should use. I used the following code:

hadimvo Openness AGGSE, generate(Hadi D) p(0.05)

Will this give me the outliers on the basis of openness? And if I use:

hadimvo AGGSE Openness, generate(Hadi2 D2) p(0.05)

Will this give me the outliers on the basis of AGGSE?
Also I decided before using this test to exclude observations that have openness>200. But according to this test values till 230 are not outliers. Should I thus leave these observations in my data?

Thanks a lot in advance,

Bor Kemkes

[Dave Airey]

hadimvo is no longer a supported Stata command. You might look at this alternative (https://www.stata-journal.com/articl...article=st0197).

[Bor Kemkes]

Hi mr Airey,

I am using stata 15.1 and I can still use the command. I understand that it is not officially supported anymore but it remains working.
This means that I can still use it right?
Thank you for providing me with an alternative.

Bor

[Dave Airey]

You are correct.

[Chen Samulsion]

hadimvo is no longer a supported Stata command. You might look at this alternative (https://www.stata-journal.com/articl...article=st0197).

I want to know why -hadimvo- are no longer supported by Stata? Or put it correctly, why it no longer an official part of Stata?

[Dave Airey]

Both bacon and hadimvo are by the same author (Hadi) and bacon is stated to be improved. Hadimvo will continue to work in Stata, just under "previously documented" with no further development.

[Bruce Weaver]

I believe that Sylvain Weber programmed -bacon- (SJ).

. net sj 10-3 st0197

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------
package st0197 from http://www.stata-journal.com/software/sj10-3
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------

TITLE
      SJ10-3 st0197.  bacon: An effective way to detect outliers...

DESCRIPTION/AUTHOR(S)
      bacon: An effective way to detect outliers in multivariate
        data using Stata (and Mata)
      by Sylvain Weber, University of Geneva,
           Department of Economics, 40 Boulevard du Pont-d'Arve,
           1205 Gen�ve, Switzerland
      Support:  [email protected]
      After installation, type help bacon
      This also requires the moremata Mata library; type
        ssc install moremata and mata: mata mlib index

INSTALLATION FILES                                  (click here to install)
      st0197/bacon.ado
      st0197/bacon.hlp
      st0197/mahalanobis.mo
      st0197/mediandist.mo

[Dave Airey]

True, but I meant that Hadi was an author of the methods, not Stata bacon implementation. I could have been more clear.


Billor, N., A. S. Hadi, and P. F. Velleman. 2000. BACON: Blocked adaptive computationally efficient outlier nominators. Computational Statistics & Data Analysis 34: 279-298.