Including Customized Offers

[tingli]

Dear,

I'm currently using Ngene for a choice design but cannot find a solution from the manual and online forum. I hope you can help. I am generating two similar designs, one of which is a simple fractional factorial design with two alternatives, 6 attributes of 2 levels each.

In another design, I need to present a customized alternative among the two alternatives, BUT ONLY in some choice sets but not in others. This customized alternative is based on respondents' answers from earlier questions. So it's different for each individual. However, the other alternative is still based on the design and do not relate to the customized alternative. The idea is to examine the effect of introducing customized offers.

Does Ngene allow me to do this? It’s similar to a reference alternative or a status quo alternative. However, we don't want to show it all the time and the other alternative is not linked to the reference. So it's different from the pivot designs explained in the manual. Can you suggest me a way to generate this? Thank you.

Ting

[Michiel Bliemer]

You can only optimize on the design efficiency once the customized levels of the alternative are known. For example, if the customized alternative constaints the levels [0,1,1,0,0,1], then the following syntax would optimize for the design (note that you have to use the correct priors and levels):

Code: Select all

design
;alts = alt1, alt2
;rows = 8
;eff = (mnl, d)
;model:
U(alt1) = b1[0]*x1[0,1] + b2[0]*x2[0,1] + b3[0]*x3[0,1] + b4[0]*x4[0,1] + b5[0]*x5[0,1] + b6[0]*x6[0,1] /
U(alt2) = b1*x1a[0] + b2*x2a[1] + b3*x3a[1] + b4*x4a[0] + b5*x5a[0] + b6*x6a[1]
$


So you can generate an efficient design for each combination. If you are able to do a 2-stage procedure, you could first ask the respondent for their status quo, giving you the levels, and then optimize a design for each respondent. Or, you can already create a database of efficient designs for all possible status quo combinations, there are 2^6 = 64 combinations possible, so that is doable.

Michiel