choice-metrics.com

Dear Ngene team,

Thank you for providing such a good users' manual and for this forum where finding highly useful answers to additional questions.

I am writing to you because I am preparing the experimental design for a study on conservation of traditional pig breeds. As in many environmental applications, I have a status quo alternative that is maintained fixed across choice sets (rows) and across respondents. I used to be quite fond of pivot designs and I have used them in the past, conducting beforehand some pilot test to get prior estimates.

However, inspecting carefully this forum I came across several posts where you provide other Ngene users with alternative ways of specifying the status quo and allowing as well for dummy of effects coding to be explicit in the design. As this will be my case (most of my attributes will be effects-coded), I think I cannot include this specification in a pivot design. Is that the case?

Hence, I was wondering what would be the advantages of using pivot designs over specifying the SQ alternative provided that the SQ alternative will be kept constant across the whole experiment. Hope you can give me some hint revolving around this issue.

Many thanks in advance for your time and advice
elsa

The choice for a fixed status quo alternative or a personalised reference alternative depends on the study. If the status quo is the same for each respondent, then using a fixed alternative is sufficient. However, if the reference alternative is different for different respondents, then a pivot design would be more useful as you do not want to present a status quo alternative that does not make sense to certain respondents. The specifications of the utility functions are essentially the same.

Dummy/effects coding is not necessarily problematic, it depends on what your utility functions look like. As long as all coefficients are identifiable in your discrete choice model, then it will work. Maybe I can provide more specific comments once I see the syntax of what you are trying to do.

PS: It is best to post general questions regarding experimental design that are not related to Ngene in the other forum.

Thank you for your answer, Michiel, and sorry for posting my question in the wrong place. Once I have the final syntax I may post it for your feedback.
e