On Tue, 22 Jul 2014, "Juan C. Estévez" wrote:
Only a precision: I click now unintentionally the arrow before
defining the
dependent variable and ... Sven is right (!?).
Thanks, Juan; this is now fixed in CVS. The problem was not specific to
panel models, it was -- as you noted -- specific to the case where the
dependent variable is not yet selected.
The thing is that when adding a variable to the list of regressors, we
first check that it's not the same variable that is selected as the
regressand. In this special case we got a spurious apparent conflict,
since the empty dependent variable slot gave an "ID number" of -1, and the
list to be added, not being a single series, also gave an "ID number" of
-1. Easily fixed.
Allin