Yes, that's the problem. The three observations for one choice set differ by an experimental design. Maybe this would be something that should be added to gretl since many economists work with this type of problem.
On 5/17/2010 8:29 AM, Sven Schreiber wrote: > Data Analytics Corp. schrieb: > >> Hi Sven, >> >> Thanks for the advice, but the problem for a stated preference discrete >> choice model is that estimation depends on the choice sets seen by the >> individuals. Say each individual sees a choice set with three >> alternatives. The estimation routine needs to know the three and which >> of the three was selected. STATA allows this. I don't see that here. >> >> > Well if the three alternatives are different from one observation to the > next (why are they always three then, though?), this is beyond me, > sorry. Probably not doable in gretl (apart from programming it up > yourself with gretl's scripting language, of course...). > > good luck, > sven > > -- ________________________ Walter R. Paczkowski, Ph.D. Data Analytics Corp. 44 Hamilton Lane Plainsboro, NJ 08536 ________________________ (V) 609-936-8999 (F) 609-936-3733 walt(a)dataanalyticscorp.com www.dataanalyticscorp.com