On Fri, 19 Sep 2008, Mariusz Doszyń wrote: > To estimate individual effects shoud I sum up (common) constant > and individual effects? No, the values returned by $ahat (or "Save/Per-unit intercepts" in the GUI model window) are the estimates of the intercept for each unit or group, nothing should be added to them.

On Fri, 19 Sep 2008, Allin Cottrell wrote: > Sorry, but I'm now thinking there might be a slight problem with > computing these $ahat values. I'll investigate further and post > again. No, apologies for the noise: the $ahat values are correct. But while I'm at it, here's a little script which shows what's going on. <script> open greene14_1.gdt logs C Q PF LF genr unitdum # ols with generic const + 5 dummies ols l_C 0 l_Q l_PF l_LF du_2 du_3 du_4 du_5 du_6 -q matrix a = ones(6,1) * $coeff[1] + (0 | $coeff[5:9,]) printf "\n%f\n", a # OLS with all 6 unit dummies ols l_C l_Q l_PF l_LF du_1 du_2 du_3 du_4 du_5 du_6 -q matrix b = $coeff[4:rows($coeff),] printf "\n%f\n", b printf "bbar = %f\n", meanc(b) # panel, Fixed Effects panel l_C 0 l_Q l_PF l_LF -q printf "F.E. const = %f\n", $coeff(0) series apanel = $ahat smpl year=1970 -r print apanel -o </script> Allin. _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users __________ NOD32 Informacje 3457 (20080919) __________ Wiadomosc zostala sprawdzona przez System Antywirusowy NOD32 http://www.nod32.com lub http://www.nod32.pl