Have an idea?

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

Literature for ex-post segmentation outside of Sawtooth

Sorry to follow up on this thread. I hope you are able to help.
I have been provided with a file of HB part-utilities at the individual level (ZC diffs).
I would like to explore segmentation by a number of dependent variables captured in another file.
Can these be used to split the sample in a separate programme such as SPSS or Stata, or must segmentation criteria be applied within Sawtooth?
To clarify, let's say there are 1000 responses. If I were to subgroup the individual part-worths by calculating the means for each segment men (n=300) / women (n=700) outside of Sawtooth, is this a valid approach?
If so, would it be possible to point me in the direction of any technical literature on the topic?

(Lastly, if possible to describe the stepwise simulation approach for the marginal likelihood or sensitivity of each level that can be contrasted between segments, that would be very helpful - and many thanks.)
related to an answer for: Approach for segmentation for CBC/HB
asked May 25 by Jules K (120 points)

1 Answer

0 votes
Sure, you can analyze part-worth utilities that have been normalized (zero-centered diffs) in any 3rd party software of your choice.  You are just doing tests on differences between means.  You can compare groups of people (e.g. old people vs. young people) on the part-worth utilities.

T-test is often used for comparing two groups.  F-test for comparing 3 or more groups.

Regarding your last question, I'm sorry but I don't know what you're referring to on "stepwise simulation approach".
answered May 25 by Bryan Orme Platinum Sawtooth Software, Inc. (201,165 points)