Page 1 of 1

orthogonal design

PostPosted: Thu Jun 09, 2011 1:16 am
by spqr83
Hi
I have a silly question. Are there any reasons why "orth" command (sim, seq, seq2 or ood) generates always the fist choice in the fist choice situation equal to the each lower attribute levels? Can someone give me any suggestions to remove this problem? Here you may see the design sintax.
Code: Select all
design
;alts=P1*, P2*, SQ
;rows=16
;orth=seq2
;block=2
;model:
U(P1) =  b0
        +b1.effect[0|0|0]*view[-3,-1,1,3] 
        +b2*co[5.5,7,8.5,10]             
        +b3*bio[-50,-25,0,10]
        +b4*land[-10,-5,0,2]
        +b5.effect[0|0|0]*recr[-3,-1,1,3]
        +b6*cost[0,25,50,75,100,125,150,175]/
U(P2) =  b1.effect*view
        +b2*co
        +b3*bio
        +b4*land
        +b5.effect*recr
        +b6*cost$

Thanks

Re: orthogonal design

PostPosted: Fri Jun 10, 2011 3:29 pm
by Andrew Collins
Hi

For most orthogonal design specifications, only a single design is returned, and the problem of having a choice task composed entirely of the lowest levels is an unfortunate property of the design that was found.

There is a solution however. Add the following property:
Code: Select all
;eff=(mnl,d)


This will search for an efficient, orthogonal design. While you haven't specified priors, the search will return more than one design, and these other designs mostly don't have the problem choice task that you encountered.

Andrew

Re: orthogonal design

PostPosted: Fri Jun 10, 2011 10:16 pm
by spqr83
Thanks a lot Andrew
Basically, there are no priors for the issues I'm evaluating, hence I'm trying to run a pilot before deliver the questionnaires to the respondents (F2F survey, very costly!). I have already tried to use "eff = (mnl, d)" command adding either zero priors or very low priors with signs (+ or - 0.000001), anyway I had some advises to use orthogonal design in the pilot. This is due, because at the moment there are no publications reporting similar dummy attributes (view and recr) and we are not sure even about the signs. If you have any other advise for removing the first alternative in the first choice situation in the orthogonal design, let me know, on the contrary I'll use "eff" command with zero priors.
Thanks again
Enrico