Choice task with two "fixed" alternatives

[andreb84]

Hi all,
I'm quite new to choice experiments, that's the reason for my question.
I am designing a choice experiment, and my idea is to include 3 alternatives (one is the status quo) in each choice task.
I was thinking if it make sense to do in this way: one alternative is the status quo, one alternative, fixed among all respondents, in which all attributes' levels are equal to 0 (including payment, this alternative refer to no-existence of the public good), and one alternative with varying levels of attributes. Does the inclusion of the second fixed alternative make sense?
Thank you

[Michiel Bliemer]

It is unusual, but it is possible and the model parameters can be estimated. For example, in Ngene syntax it could be something like the following:

Code: Select all

design
;alts = sq, fixed, varying
;rows = 12
;eff = (mnl,d)
;model:
U(sq)      = b1 * Asq[1]    + b2 * Bsq[2] /
U(fixed)   = b1 * Afixed[0] + b2 * Bfixed[0] /
U(varying) = b1 * A[0,1,2,3]  + b2 * B[0,1,2,3]
$

As long as people are trading off between the status quo and fixed alternative, then I think it makes sense. If the status quo is better/worse than the fixed alternative, then it does not make sense as one of the alternatives is dominated by the other.

Michiel