Have an idea?

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

Zero-Cent. Utilities between segments not compareable

Relating to what you've just been explaining, I am using the zero-centered diffs method to get my utilities. Generating two segments (by nationalities) via the "Utility report" based on my HB-model I've got for instances the following utilities

                                        Segm. 1    Segm. 2
Attribut level 1         30,85         23,96        
Attribut level 2         10,08         -2,61
Attribut level 3         -7,02          -6,96
Attribut level 4         -33,91       -14,38

So, the positive sum of segment 1 would be 40,93 while the positive sum of segment 2 (out of the exact same survey and data) is 23,96.
Is this due to the fact the my total sample contains 3/4 segment 1 comparing to 1/4 segement 2?

Is there any chance how I could still compare the 2 segments?

Thanks in advance,

asked Sep 25, 2018 by Ben

1 Answer

0 votes

Those values are means, based on the respondent level data.  You can compare them most easily by a t-test, as you would for any other metric variable (e.g. a count, a rating scale).
answered Sep 25, 2018 by Keith Chrzan Platinum Sawtooth Software, Inc. (103,225 points)