Bookmark and Share

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


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Omitted category interaction variables


From   Maarten Buis <[email protected]>
To   [email protected]
Subject   Re: st: Omitted category interaction variables
Date   Mon, 4 Nov 2013 15:44:20 +0100

OK, I suspect that at the point where the factor variables are parsed
Stata does not "know" or "realize" that the main effects are included
via the fixed effects, so it "thinks" that none should be excluded. At
the estimation phase Stata "realizes" that there is perfect
multicolinearity and it excludes one. This automatic exclusion is not
very consistent, and there is your problem. I think that what you want
to do is reasonable and that this behavior of Stata is undesirable.
However, I don't know how easy it would be to fix, as I suspect that
the parsing of factor variables is done in separate programs outside
the estimation command while the fixed effects are a property of the
estimation command.

You could go manual and create the indicator (dummy) variables and the
interaction terms yourself, thus avoiding the whole problem.

-- Maarten


On Mon, Nov 4, 2013 at 2:58 PM, Estrella Gomez <[email protected]> wrote:
> The thing is I cannot include the main effects because these vary over
> pair of countries, and I'm including country pair fixed effects in the
> estimation. When I do not include the main effects, Stata excludes one
> year by default, and it is not always the same. I would need it to be
> the same because I want to compare the evolution of ldist, dhome and
> dlang over time
>
> Thanks a lot,
> Estrella
>
> 2013/11/4 Maarten Buis <[email protected]>:
>> On Mon, Nov 4, 2013 at 2:18 PM, Estrella Gomez wrote:
>>> Yes, but this is not what I want. Then it excludes the year 2006  in
>>> the last set of dummies; i.year; but I want it to be excluded from
>>> every interaction term (i.year#c.ldist; i.year#i.dhome and
>>> i.year#i.dlang)
>>
>> You forget to include the main effects of ldist dhome and dlang. That
>> is probably what is causing the problems, as in that case you don't
>> want to exclude a reference category. If you include the main effects,
>> then setting the baseline once will be enough.
>>
>> -- Maarten
>>
>> ---------------------------------
>> Maarten L. Buis
>> WZB
>> Reichpietschufer 50
>> 10785 Berlin
>> Germany
>>
>> http://www.maartenbuis.nl
>> ---------------------------------
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/



-- 
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

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


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index