st: Varying coefficients with -mvreg-

["Ro'i Zultan" <zultan@bgu.ac.il>]

Dear statalist,

The -mvreg- help says "The individual coefficients and standard errors produced
by mvreg are identical to those that would
be produced by regress estimating each equation separately."
However, with some variables, I get a different coefficient for one dependent
variable (which is the same as I get with -regress-) and for the same
variable as part of two or more dependent variables. E.g.,
reg q1 pol
and
mvreg q1 = q8
provide the same results. However, the coefficient and standard error for
q1 is different with
mvreg q1 q2 = q8
or
mvreg q1 q2 q3 = q8
With other independent variables, the coefficient is always the same. Where
does this difference come from?

Thanks,
Ro'i
*
*   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/