hi all
i am working with an unbalanced panel data set, and i encounter the
following problem :
there are two ways to estimate the fixed and between estimators. the first
is to use the commands "xtreg,fe" and "xtreg, be", while the second is to
calculate within-group averages, and deviations thereof, of the right hand
side variables, and then use these averages and deviations in a "xtreg,re"
specification, so that the coefficient on the averages are the "be"
coefficients and the coefficients on the deviations are the "fe"
coefficients (this second method is described in
http://www.stata.com/support/faqs/stat/xt.html).
the problem i have is that these two methods are giving me very different
answers, especially for the between regressions.
can anyone tell me why i am getting such different results, and which one i
should trust? thanks,
