Have an idea?

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

ACBC segmentation

I want to do segmentation based on my ACBC survey data. Previously, I was informed that I can use CCEA for it. However, I am using the software with academic license from university computers and it is not really possible for me to download and install CCEA, even though it is free. It requires too many authorisation and I will not be able to get that. In that reason, I would like to is there any other way for me to do that? is it possible to convert the ACBC to use on Sawtooth Latent Class for segmentation?

Thank you !
asked Dec 9, 2014 by Mir (400 points)

1 Answer

0 votes
That is too bad that you cannot get authorization to install CCEA.

Using our latent class MNL approach with latent class data is problematic, because a key section that drives the utilities is the "Screener" section, which involves indicating for many product concepts whether they are a possibility or not.  The "or not" choice is treated like a None choice, leading to a "None Utility."  The more times a respondent has a tendency to choose "not a possibility", the larger the "None utility" for ACBC.  This can become a major driver of the final segmentation solution that latent class finds for ACBC data, which may not be terribly useful--because the tendency to report that concepts are "not a possibility" may be due to response style bias for individuals.  For example, some people are "yea-sayers" and some people are "nay-sayers."

Maybe there is a software system that your firm allows you to already have or install that has cluster analysis?  SPSS, SAS, R, etc. each have cluster capabilities.  Although you cannot do the full ensemble analysis recommendation I gave you in a previous post, you could use these cluster tools to develop cluster solutions from the ACBC part-worth utilities (the zero-centered diffs, normalized utilities).
answered Dec 9, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (174,440 points)