Hello Sawtooth community,
I am new to the topic of conjoint designs and analyses and probably don't have a usual conjoint research question, as we are conducting a study on transport choice.
Nevertheless, I think that a conjoint fits well with our research question; I just don't know how to structure the design efficiently yet, and I hope for your input.
First of all, the study:
We want to investigate how certain factors (transportation alternatives, directness of the alternative, directness of the e-scooter, peers, time of day) influence the decision for or against an e-scooter. (Our focus is on night-time trips by young people).
This means that in our decision-making situation, an e-scooter is always available in addition to the alternative means of transportation.
To make things more appealing and immersive for the young target group, we generate image scenarios at night in which the factors of interest are varied.
Our minimum N = 300 respondents.
I am working with Lighthouse Studio 9.15.6 for Academic Research.
Idea 1 – CVA
CVA with single concept tasks, 10 versions with 9 tasks per version, see example bellow
Advantages: We can visualize a situation graphically, which comes close to a decision for a transport alternative in reality; the D-efficiency is close to 1.0 for all attributes and versions
Problem 1: As I understand it, with a CVA, we cannot (or only to a limited extent) benefit from the great analysing options that Lighthouse provides, but that would be important to us, e.g. the simulator
Problem 2: We are also interested in the interactions of the attributes and, as I understand it, CVA is only suitable for this to a limited extent
Idea 2 – CBC
·        CBC – discrete choice with 12 random tasks and 2 concepts per task, see example bellow
Advantages: All CBC analyzing options would be available to us, and interactions can be studied
Problem 1: Being confronted with two situations seems very artificial to me (the image on the left is still missing in the example), as I can, of course, only be in one situation in reality
Problem 2: I fear a loss of information because, in contrast to the CVA, I am not measuring a quasi-metric outcome (scale), but only a nominal one (1/0)
Idea 3 – CBC modified
CBC as above as a dual concept task
Changing the attributes so that e-scooters are included as one of the transportation alternatives -> then only allow concept combinations via alternative-specific rules where one is an e-scooter and the same other levels appear in each case (e.g. same time in both cards)
Aim: Presentation of only one image with the question: Which mode of transportation would you choose in this situation?
Advantages: It would be a realistic decision situation for the respondent, as with the CVA, but with the evaluation advantages of the CBC design
Problem: Of course, many rules and conditional relationships between the attribute (levels) would have to be inserted, which would certainly reduce the efficiency of the design
To make things even more complex, I would like to ask follow-up questions if the respondent chooses the e-scooter, e.g. How likely would you be in this situation to choose the e-scooter if you were under the influence of alcohol? with e.g. a likert-scale as response format.
I have not yet succeeded in implementing this in either the CVA or the CBC.
Sorry for the long post!
I welcome all comments, tips and corrections (in case I have misrepresented something) 😊.
Best wishes,
Juliane