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

# st: Testing the difference of marginal effects in Stata

 From "Mongeon, Kevin" To "'statalist@hsphsun2.harvard.edu'" Subject st: Testing the difference of marginal effects in Stata Date Thu, 28 Apr 2011 15:02:44 +0000

```Hi:  I am a beginner at Stata so I apologize if my question is elementary.

I am running a logit and probit that contains a number of indicator variabl= es.  I would like to test whether the marginal effects of the indicator var= iables are statistically different (one sided) from each other.

Here is some of my code:

Logit y constant up1 dn1 up2 dn2
margins, dydx(*) post

up1 dn1 up2 dn2 are all indictor variables.  I have other control variables=  so I cannot drop the constant.

What I would like to do is test whether the marginal effects of up1 is grea= ter than dn1.

Thanks
-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of philippe van kerm
Sent: Thursday, April 28, 2011 10:57 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: Negative incomes and income components using -sgini-

> -----Message d'origine-----
> De : owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] De la part de David Coyne
> Envoyé : Thursday, April 28, 2011 2:58 PM
> À : statalist@hsphsun2.harvard.edu
> Objet : st: Negative incomes and income components using -sgini-
>
> Hello,
>
> I've been attempting to decompose the Gini coefficient of a measure C,
> where C=A-B. I have two questions related to this. First, if C has some
> values that are negative, how does this affect the interpretation of
> the
> Gini coefficient?

Negative data in relative inequality measures (such as the Gini) is a source of concern. Relative inequality measures after all summarize the distribution of shares of a total -- how should we interpret negative shares?

Mechanically however, Gini coefficients can be computed with negative data -- at least as long as the mean is different from zero (unlike measures relying on log transformations, such as Mean Log Deviation or Theil indices).

You may consider absolute inequality measures if negative data form a significant part of your variable of interest. However, I am not sure how to proceed with the decomposition by source in this case. It is very likely do-able but  -sgini- does not implement this (I should give this a go one day).

> Second, if B is nonnegative, (i.e. using -sgini-
> would
> have me use C=A + (-B), so if (-B) < 0 for all observations) thus
> leaving
> me with a negative contribution (s*g*r) to the Gini coefficient of C,
> does
> that mean that B has an equalizing effect on C? Not sure of the
> interpretation of this. Any insight and/or advice would be appreciated!

If all your data are negative on -B, then I think you can safely interpret a negative contribution measured by s*r*g in -sgini- output as indicative that B has equalizing impacts.

Philippe

>
> --
> David Coyne
> *
> *   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/

Attention, nouvelle adresse a partir du 15 avril 2011 :
3, avenue de la Fonte
L-4364 Esch-sur-Alzette

*
*   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/

*
*   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/
```