Have an idea?

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

Combining individual utilities with segmentation variables in an independent sample t-test in SPSS

In my CBC I included extra segmentation questions concerning consumers' attitudes regarding the dependent variable.
Now I was wondering whether it is allowed to combine the individual NONE utilities obtained via the CBC-HB with the segmentation question about consumers' attitudes in an independent sample t-test in SPSS?

I assume the NONE utility gives information about whether participants want (or do not want) to participate in, or buy the dependent variable, right?
asked Mar 27, 2017 by anonymous
edited Mar 27, 2017 by Walter Williams

1 Answer

0 votes
This is a bit challenging, but perhaps can be done.  

I'll assume you are using CBC with the "standard None", where None is one of the concepts in every CBC task.

Next, I'll assume you are using HB estimation such that you obtain a vector of utilities for each respondent.

I'll assume you want to remain with frequentist t-tests rather than delve into Bayesian statistical testing "on the draws."  You've said you want to to t-tests using SPSS.

Given all that, please recognize that the None utility is typically understated in marketing contexts.  Meaning: people tend to use the None alternative a lot less than they would in the real world.  People's intent to buy is thus overstated.  Still, you might want to treat None as a relative measure of purchase interest, to compare different groups.

Also, please note that you should use the zero-centered Diffs rescaled utilities in T-tests, as these are normalized to better put each person on the same part-worth utility scale.  The raw utilities can show pretty large differences in scaling between people due to respondents answering with different amounts of response error.  Lower response error leads to larger magnitude utilities across the boards, and vice-versa.
answered Mar 27, 2017 by Bryan Orme Platinum Sawtooth Software, Inc. (131,890 points)
...