Beginner questions for pilot DCE design

[Nerida]

Hello Michiel, John and Andrew,

This is my first DCE and I’m also trying to teach myself to use Ngene. I’m attempting to develop a simultaneous orthogonal design of 36 rows in four blocks. I’ve five attributes with four levels and one attribute with two levels. My questions are basic but I'm struggling to find the information I am looking for elsewhere in the forum. I hope you can help me.

The syntax I have used is:

Code: Select all

? This will generate a simultaneous orthogonal factorial design with four blocks
Design
;alts = alt1, alt2
;rows = 36
;orth = sim
;block = 4
;model:
U(alt1) = b1 + b2 * A[0,1,2,3] + b3 * B[4,5,6,7] + b4 * C[8,9,10,11] + b5 * D[12,13,14,15] + b6 * E[16,17,18,19] + b7 * F[20,21]  /
U(alt2) = b2 * A          + b3 * B          + b4 * C            + b5 * D              + b6 * E              +  b7 * F $


A design for a pilot trial has been generated, and from what I can tell it looks fine.

Question for the experts #1: Can I confirm that it is as simple as using the syntax above? Have I overlooked any important steps/details?

The first choice set of the 36 rows generated is not a feasible option: Both options contain the same combination of attribute levels. I need to remove it. After reading the manual and several forum posts, I get the impression that simply removing this choice set will affect orthogonality, but improve realism for respondants and therefore is best option. I tried to integrate a constraint into the syntax to remove the non-feasible option but it did not work, and no design was generated.

Question for the experts #2: Can I confirm that to remove a non-feasible choice set, one simply removes it when converting the design into the questionnaire? I can’t find any instruction in the manual how to remove and replace non-feasible choices, and wonder if I have missed it somewhere?

Any advice or direction is appreciated.

Thank you for your time,

Nerida

[Michiel Bliemer]

Hi Nerida,

1. The syntax is fine, but you need to make a distinction between labelled and unlabelled experiments. Your utility functions indicate an unlabelled experiment where both alt1 and alt2 are generic alternatives of the same type. In that case, you should not use a simultaneous orthogonal design, which is only for labelled experiments in which attributes typically and different attributes and coefficients, but rather a sequential orthogonal design. So you can use:

;orth = seq

This avoids the issue that you discussed, choice tasks should now no longer be the same across alt1 and alt2.

2. Yes you simply remove the row in the design before you implement it into the survey instrument, i.e. you remove one choice task from the questionnaire.

Note that Ngene related questions should be asked in the Ngene forum, while non-Ngene related questions can be asked here in the stated choice experimental design forum.

best wishes,
Michiel

[Nerida]

Michiel, thank you for your response and I'm really sorry about posting in the wrong forum. See, I've got no idea what I'm doing?!!

I appreciate the advice and I'll implement the changes.

Kind Regards,

Nerida