Re: st: test for equality of coeff. across models with different samples

[Nick Cox <njcoxstata@gmail.com>]

I should have said "the main issue here".

On Tue, Apr 3, 2012 at 12:07 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> I don't think that is the issue here. The issue is that one model has
> fixed effects and the other doesn't. Whatever the best answer is, it
> is not using -test-, as -test- does not compare those two models; it
> just works on the model in memory.
>
> Nick
>
> On Tue, Apr 3, 2012 at 11:55 AM, David Hoaglin <dchoaglin@gmail.com> wrote:
>> Bettina,
>>
>> If one sample is a subsample of the other, the two estimates of a
>> particular coefficient are not independent.  You can "test their
>> equality" by fitting a model to the full sample that includes an
>> interaction between the subsample and each predictor (for the
>> constant, an indicator for being in the subsample).  The test would
>> then be that the coefficient of the interaction is zero.
>>
>> David Hoaglin
>>
>>
>>> I estimate two models and want to test for equality of coefficients across the equations. At the same time the samples I use differ across models (one is a subsample of the other) .
>>> More specifically I estimate:
>>> Logit dep.var independent vars
>>> Logit dep.var independents vars, fe
>>>
>>> I want to know whether coefficientlist_eq.1=coefficientlist_eq.2 foreach independent variable.
>>> Is test the appropriate command here?
>> *
>> *   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/