Have an idea?

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

Heterogeneity in CBC/ HB - How to evaluate it precisely?


I run an HB analysis to stress that consumers in my research context have highly heterogeneous preferences, i.e. managers should serve their (potential) customers individually. I've seen an interesting post by Sawtooth regarding the issue to account for heterogeneity: https://sawtoothsoftware.com/forum/10485/heterogeneity-in-cbc-hb?show=10485#q10485

However, when thinking about standard deviations (SD), I am not sure how to use them within CBC/HB to represent heterogeneity. I guess you can only compare the SD among one attribute slthough one looks at zero centered diffs, don't you? If so, what is the "reference value" one compares to?

Since the heterogeneity of preferences is the one of the main topics of my academic research, it would be great to get some further advice for looking at certain values, computing and representing them. It would be great to hear from you soon.

Best regards! :-)
asked Jul 22, 2017 by briniminii (490 points)
retagged Jul 23, 2017 by Walter Williams

1 Answer

0 votes
Best answer
You could express the distributions of utilities for a given level of a given attributes using standard measures (standard deviation, skewness, kurtosis) but I think more often people just look at the distributions:  are they unipolar or bipolar; are they tight on the mean or spread; are they skewed in one direction?  I don't usually see a lot of quantitative analysis of the heterogeneity.  Rather that heterogeneity contributes to learnings from the research, by providing the means for looking at different groups of respondents (segmentation) or by providing more accurate simulations, where the tastes of unusual customers aren't lost in the mean.
answered Jul 23, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (65,925 points)
selected Jul 23, 2017 by briniminii