choice-metrics.com

Via Ngene, we created the questionnaire of 2 levels of 9 attributes plus 3 levels of 1 attribute by the following code.

/*** --- Code --- ***/
Design
;alts = alt1, alt2
;rows = 24
;orth = seq
;model:
U(alt1) = b0 + b1 * A[0,1] + b2 * B[0,1] + b3 * C[0,1] + b4 * D[0,1] +b5 * E[0,1] + b6 * F[0,1] + b7 * G[0,1] + b8 * H[0,1] + b9 * I[0,1] + b10 * J[0,1,2]/
U(alt2) = b1 * A + b2 * B + b3 * C + b4 * D +b5 * E + b6 * F + b7 * G + b8 * H + b9 * I + b10 * J
$
/*** --- End of Line --- ***/

Every time we run it, the combination of attribute A stays the same, but that of attribute B fluctuates. Is there any way to make the combination of an option remain unchanged by the same code?
We tried to fix the seed by using ‘rseed’, but it did not work, and hence, the combination of attribute B still changed.
Furthermore, is it possible to make the combination of attribute A change at each run?

Sequential orthogonal designs are created as follows:

1. Find an orthogonal design for Alternative A
2. Create profiles for Allternative B based on a randomisation of the order of the rows for Alternative B

Because of this randomisation, the resulting design is different each time, but the rows used in B are identical, they just appear in a different order.

If you want the exact same design, you should specify that you want a specific sequential orthogonal design, namely a D-optimal orthogonal design. You can do this by replacing ;orth = seq with ;orth = ood. This design maximises trade-offs between the two alternatives and has no random component.

Michiel

Thanks so much for your quick response and your comments.
I have one more question.
Is it possible to use a seed or something so that the combination of attribute A changes each time it is executed?

No there is currently no such random seed option available, only for Bayesian draws and random draws there are ;bseed and ;rseed.
It is a good suggestion though so we may add such an option in a next version.

Michiel