Statalist The Stata Listserver


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

st: RE: RE: significance of grouped contributions in Oaxaca decomposition


From   "Van de Poel, Ellen" <VandePoelE@who.int>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: significance of grouped contributions in Oaxaca decomposition
Date   Mon, 12 Jun 2006 13:54:27 +0200

Dear Ben,

Thanks a lot for your help!

Ellen

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Ben Jann
Sent: 09 June 2006 23:48
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: significance of grouped contributions in Oaxaca
decomposition

Ellen Van de Poel wrote:
>  I use the command : oaxaca group1 group2, de esave
> 
>  to generate detailed Oaxaca decomposition results.
> 
>  Thereafter I want to group the contributions of some binary variables
>  that result from a categorical one. E.g. I want to group
contributions
>  from all the (binary) income quintiles to get a general contribution
> for income.
>  To get the statistal significance of this grouped contribution I use
>  the command :
> 
>  nlcom (wealthcon:
> _b[quintile1]+_b[quintile2]+_b[quintile3]+_b[qquintile4])
> 
>  Now, the problem is that this only works for the "endowments
>  contribution". Through the _b[] command I cannot seem to get the
>  contributions from the coefficient and interaction effects.

There are two issues:

First, and most important, -nlcom- will not produce correct results
concerning "statistal significance" in any case. Here is the description

of the -esave- option from the -oaxaca- help file:

esave specifies that the results be returnd in e(). This is useful,
     e.g., if you want to use -bootstrap- with oaxaca. Note that the
     off-diagonal elements in e(V) will be set to zero since oaxaca does
     not provide the covariances among the various decomposition
     components. Do not apply -lincom- or similar techniques to the
     returned results.  Also do not use -predict-.

Clearly, -nlcom- is one of these "similar techniques". The reason why 
-nlcom- or -lincom- do not compute correct results is that -oaxaca- 
just sets all covariances in e(V) to zero. This, in fact, is also the 
reason why -oaxaca- does not return its results in e() by default. I 
just added the -esave- option to facilitate tasks such as 
bootstraping.

The correct way to compute combined effects is to group variables 
using the detail option. For example, Ellen could type

 . oaxaca group1 group2, detail(quintile=quintile*)

The contributions of all the quintile variables would then be combined
internally and the correct standard error will be reported.


Second, the results stored by -oaxaca- in e(b) and e(V) are organized
in several "equations". For example, -oaxaca group1 group2, de esave-
will save three equations: "endowments", "coefficients", and 
"interaction". Ellen can access the the results for the different 
equations as follows:

 endowments equation: _b[quintile1] or [endowments]_b[quintile1]
 coefficients equation: [coefficients]_b[quintile1]
 interaction equation: [interaction]_b[quintile1]


ben

*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index