Hello Guys,
I am doing my first conjoint project (pairwise CVA) and I'm trying to do hiararchical cluster analysis based on conjoint results.
I tried to run hiererchical cluster analysis on logit recoded utilities and zero-centered utilities (SSI Web - OLS - Recode Method). The problem is that I get totally different segments when based on logit in comparison to zero-centered utilities. I expected some differences, but I did not expect so much difference (caused only by different recode method: logit vs zero-centered).
Is there any "best practice" saying that for clustering one should use logit or zero-centered utilities? How can I decide which solution is better in my case?
Many thanks in advance.
Cluster analysis based on CVA utilities
Resolved
1 reply