Avoiding dominant alternatives

[Anat Tchetchik]

Dear all,
I hope this is not too trivial question (I have search the manual and the forum and could not find the answer)
I am trying to create a very simple generic orthogonal design with 3 alts, each is described by the same 3 atts, each att has the same 3(ordinal categorical) levels.
this is the code:

Code: Select all

Design
;alts = alt1, alt2, alt3
;rows = 12
;orth = sim
;model:
U(alt1) = b1 [0.00001]* A[1,2,3] + b2[0.00001]*  B[1,2,3] + b3 [0.00001]* C[1,2,3] /
U(alt2) = b1 * A+ b2 * B + b3*C/
U(alt3)= b1 * A+ b2 * B + b3*C$

I don't have priors, yet all the atts. are positive ones, so I assigned very small positive prior values.
The problem is that some scenarios includes dominant (or inferior) alternatives that can yield no info.
for example:

Scenario 3
alt1 alt2 alt3
a 3 1 1
b 3 2 2
c 3 1 3

I have also tried the option ;orth = ood, yet the problem remains.
How can I avoid these situations?
Any help on this will be highly appreciated!

[Michiel Bliemer]

Orthogonal designs ignore priors and cannot avoid dominant alternatives.
Only efficient designs can avoid dominant alternatives. All you need to do is replace ;orth = sum with ;eff = (mnl,d) and replace ;alts = alt1, alt2, alt3 with ;alts = alt1*, alt2*, alt3*.
Note that it may be that there does not exist any design without any dominant alternatives. In that case you will have to increase the number of levels of each attribute.

[Anat Tchetchik]

Dear Michiel,
Thank you very much!
This has solved the problem
Best,
Anat