Have an idea?

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

Can I use Sawtooth results to inform the creation of value functions and weights?

I conduct systems engineering trade-off analyses using multiple objective value analysis (MOVA) approaches as described by Ralph Keeney's "Value Focused Thinking: A Path to Creative Decision Making," Craig Kirkwood's "Strategic Decision Making:  Multiple Objective Decision Analysis with Spreadsheets," and Greg Parnell's "Decision Making in Systems Engineering and Management."   I am hoping to use conjoint analysis to generate value functions and weights associated with multiple objective value analysis.  Is this indeed a legitimate use of conjoint analysis output or are there nuances that may cause one to pause before implementing such an idea?
asked Mar 8, 2019 by mvcilli (120 points)

1 Answer

0 votes
I am not familiar with those books, but I have seen folks use conjoint analysis to develop weights for supporting decisions in multi-attribute contexts.  It seems like a natural to me (in fact, the first book on conjoint analysis was titled "Analyzing Decision Making" by Jordan Louviere).
answered Mar 8, 2019 by Keith Chrzan Platinum Sawtooth Software, Inc. (92,075 points)
I should clarify and say that I've seen folks use conjoint analysis (and MaxDiff/best-worst scaling) to develop importance weights for use in linear models, say where they want a set of value weights to multiply by some set of performance scores, and where they sum the corresponding products to come up with an overall score.
Thanks, Keith.  I share your suspicion that the MOVA and Conjoint complement each other well but I haven't seen it discussed explicitly in the literature.    I am about to embark on an extensive literature search to see if anyone has successfully integrated the two in practice or has perhaps conducted some research with regards to the potential limitations of such an approach.  I am hoping responses to this post could provide some leads for my literature search.  I'll of course post my findings back to this forum.
Thank you, please do.  My (slight) familiarity with MOVA suggests that it often involves linear programming, something I covered in one class back in school and that included in one academic paper in the intervening years, so I'm hardly an expert.