I have a question about the calculation of the test that all u_i=0. I am
using xtreg, fe and accidentally included a time invariant variable in the
model to be estimated. Of course Stata dropped the variable and calculated
the model. I corrected my do file and dropped the time invariant variable
but when I ran the regression again I realized that the F test was
different this time. As expected the coefficients and s.e. are the same
but not the F-test (partial output shown below). Any idea what is going
on?
Thanks,
Alejandro
- - - - - - - - -
Results of model that included the time invariant variable
Fixed-effects (within) regression Number of obs = 170
Group variable (i): codigohogar Number of groups = 85
F(12,73) = 1.42
corr(u_i, Xb) = -0.3672 Prob > F = 0.1780
---------------------------------------------------------------------------
sigma_u | 2.230205
sigma_e | 1.9891388
rho | .55694769 (fraction of variance due to u_i)
---------------------------------------------------------------------------
F test that all u_i=0: F(84, 73) = 1.36 Prob > F = 0.0922
Results of model with time invariant variable excluded
Fixed-effects (within) regression Number of obs = 170
Group variable (i): codigohogar Number of groups = 85
F(12,73) = 1.42
corr(u_i, Xb) = -0.3672 Prob > F = 0.1780
---------------------------------------------------------------------------
sigma_u | 2.230205
sigma_e | 1.9891388
rho | .55694769 (fraction of variance due to u_i)
---------------------------------------------------------------------------
F test that all u_i=0: F(84, 73) = 1.42 Prob > F = 0.0630
*
* 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/