Select Page

This session was originally presented at the Sawtooth Software European Conference in September 2020, By Steve Bell.

The question is to identify whether multiple underlying dimensions can be found to what is usually called the “importance” of attributes.
2 ways of measuring attribute importance are used in this presentation:
  • First using pair comparison (“which of these 2 attributes is more important for your choice”)
  • Then a MaxDiff exercise (“of these 5 attributes, which one is the most important, which is the least important”)

Steve reveals 2 dimensions of “importance” from an extensive set of data: based on an actual real-world survey conducted with 750 physicians and providing almost 14,000 pair comparisons from 18 rated attributes.
So through this presentation, you will get some tips on:

  • How to write product attributes that allow fair comparisons
  • What to do with attributes overlapping
  • What is the effect of considering 2 possible dimensions: impact, and scope– or generality- of attributes?
  • How to consider the difference in scope of attributes to draw the analysis, when one attribute is possibly a part of another
  • How to possibly filter out the noise from differences in scopes of attributes
  • How to calculate and present dissimilarity from pair comparisons
  • How to apply multidimensional scaling (MDS) to the data
  • How well can choice modeling interpretation predict pair comparison choices?
  • How to draw a distance-based choice model (DBCM)

Finally, comparing the maps obtained through Multidimensional scaling vs. distance-based choice modeling and their relative performances to predict pairwise choices, Steve describes ways to estimate product attribute ratings even when the attributes have very different levels of scope (or generality).

Fill in the form to get access to the video of the presentation!
Steve-Bell-AA

About our Expert, Steve Bell, Consulting Methodologist for APLUSA:

Steve is a statistician and consultant specializing in pharmaceutical markets and quantitative marketing research methods. He has applied his statistical and software development background to create several original quantitative methods tailored to the needs of pharmaceutical marketers. Steve was an Assistant Professor of Marketing at New York University. He taught courses on marketing strategy in the undergraduate, MBA, and Executive Education programs.

Steve led all advanced quantitative marketing research projects and is one of the senior methodologists at APLUSA. Steve has an MBA and Ph.D. from Columbia University. He has published articles in medical journals and in academic marketing journals. (List of publications and presentations available on request.)

Steve has also given presentations at the PMRG Research Institute, AMA Advanced Research Techniques (ART) Forum, and the Sawtooth Software Conference. (List of publications and posters available on request.) Steve is a hands-on project manager who loves innovative methodologies as well as the nuts-and-bolts of marketing research production and reporting.