On Wed, 2 Jul 2008, Gordon Hughes wrote (regarding the generation
of panel-unit dummy variables):
As a further thought I have just tested the script without the
smpl command -see the second attachment (sfa_panel2.txt). Then
the dummy variables are created properly and the script works as
intended, so there is a bug linked to the way in which the
restricted sample is handled.
OK, got it now. This should be fixed in CVS.
Allin.