# Re: st: Dummy question

 From Ernest Berkhout To statalist@hsphsun2.harvard.edu Subject Re: st: Dummy question Date Mon, 29 Nov 2004 13:53:33 +0100

As I see no responses yet, i give it a try. Quit some years ago i did something similar according to the famous Krueger&Summers (1988) article on industry wage differentials. In short, it is just taking the (weighted) average of all the dummies for a variable (say: industry) including the omitted category which has a value of zero. That is the average against which all coefficients can be measured:

gen differential = estimate - average_estimate.

But in practice it is more complicated, especially when you want to establish significance using standard errors. You need some type of adjustment there. You might want to look that up in the Krueger&Summers article.
If Javier decides to go that way i can send him -privately- my applied do-files from that time. They are however case-specific, and lack the five extra years of Stata knowledge that i have gained since then... :-)

At 23:46 28/11/2004, Javier Bacarreza wrote:

```Hi to All!!

I would like to know what should I do to be able to
interpret the dummy
variables with respect to the grand mean and not with
respect to the omitted
reference category.
```
```[...cut...]

```
```There should also be some way of recalculating the
regression results of the
dummy variables got with the n-1 dummmies to
recalculate it to the
n-dummies. How can I do that? Or what would be easier?

It's problematic, that I have 3 different types of
dummies: for regions
dum_agr, ...dum_vit (106 regions); for industries
dum_011...dum_930 (200
industries) and for time dum_2001, dum_1991 (together
3 years).

I added here one stata output that is estimated with
omitted categories (n-1
dummies) for region, industry and year dummies.

I also added one pdf-article where in the page 399 the
restrictions are
explained, but as I said I couldn't insert the
constraint to the stata.

I would like to interpret my dummies followingly: if
the region dummy
coefficient is positive and significant, the
industries located in this
region have higher than average positive location
effect of this region.

If the year dummy is positive and significant, in this
year had the higher
employment growth supportive effect than the years
together in average.

How could I calculate coefficients for n dummies in
stata to avoid perfect multicollinearity trap??

The endogenous varibale is the employment growth in
industry i located in
the region r.
```
Ernest Berkhout
SEO Amsterdam Economics
University of Amsterdam
******************************************************
For information on flexwork & research: www.tempworkresearch.nl ******************************************************
Room 3.08
1018 WB Amsterdam
The Netherlands

tel.:+ 31 20 525 1657
fax:+ 31 20 525 1686
http://www.seo.nl
===========================
A statistician: someone who insists
on being certain about uncertainty
===========================
*
* 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/