A bit of a conceptual/methodological question here.
I have a client who wants to work out how much value customers place on different aspects of an essential service. Undoubtedly most if not all of the aspects will be valued, but customers will not easily place a dollar figure on each aspect, and indeed are likely to say that most, if not all, of the aspects being tested should just be 'part of the service' and not charged for separately. Of course, the provider would never actually charge for these services separately, but needs to calculate $ value for each aspect for input into cost-benefit analyses.
There are too many aspects to test in a choice model (35 - 45), and there are no variables other than the aspects and the value attached to each of them, so I am thinking perhaps I could use a Max-Diff to work out relative value, and then create some kind of anchoring question which establishes the perceived value of a couple of key aspects (I'm thinking this will need to be the same aspects across all respondents, so I'll have to pick a couple out a-priori rather than basing them on each respondent's MaxDiff results). I can then work out average (or median) value of these aspects and from this deduce average (or median) value of the other aspects using the relativities from the MaxDiff.
Is this a good or bad idea? Are there any considerations I haven't thought of that should be taken into account?