Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Can conjoint analysis be used for portfolio decisions?

I am working my way through Bryan's excellent text on conjoint analysis, and I have a perhaps naive question. Most of the examples are framed in terms of competing product concepts defined by product attributes and levels. But I am wondering if I can use conjoint analysis to help with a portfolio research problem. Our association is considering raising dues, and we want to know the value our members place on various uses for those dues. There are four categories (attributes of the portfolio) and several options (levels) for each. For example:

Expand exisitng services
      Expand LOMA program
      Expand ethics program
      Expand advocacy program
      Don't expand exisitng programs
Launch new services
     New lawyer training
     New lawyer mentoring
     Practice support referral
    Don't launch new services
Financial objectives
    Pay down debt
    Rebuild reserves
    Balance budget
    Don't work towards financial objectives
Dues increase
    $5, $10, $15, $20

Question: Can conjoint analysis be used to inform a portfolio decision like this, or is there something intrinsic about the way CA works such that the attributes have to be about a single entity, like a product?
asked Dec 6, 2012 by anonymous
retagged Apr 21, 2014 by Walter Williams

1 Answer

+1 vote
There's no real reason that CA has to be about a discrete physical product.  CA and other choice modeling methods have been used to measure discrete products, sets of actions a firm might take (as in this case), portfolios of products (i.e. product lines), sets of advertising statements, alternative marketing or pricing strategies and so on.
answered Dec 7, 2012 by Keith Chrzan Platinum Sawtooth Software, Inc. (92,075 points)
I agree with Keith.  I wanted to comment on your attribute list, that you take care regarding the assumption that levels are mutually exclusive within each attribute.  For example, with your attribute level formation for the first attribute of:

     Expand LOMA program
      Expand ethics program
      Expand advocacy program
      Don't expand exisitng programs

You will not be able to predict interest in expanding 2 or 3 of those programs, such as LOMA & Ethics programs.  Respondents will have never seen such a profile, so the estimation of a utility corresponding to expanding 2 or 3 of those programs together is not possible.
Very helpful. Thanks so much!