FIMIX segmentation

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
fraago
PLS Junior User
Posts: 1
Joined: Wed Sep 08, 2021 2:28 pm
Real name and title: Francesca Agostoni

FIMIX segmentation

Post by fraago »

Hi,
I have conducted a Finix Segmentation Analysis using SmartPls3. I would now need to know how to do the following two things:
1. How can I see if the differences among the path coefficients among the segments are significant?
2 I want to describe the segments with other variables. how can I export the respondents allocations to segments in R?

Thank you!
jmbecker
SmartPLS Developer
Posts: 1281
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: FIMIX segmentation

Post by jmbecker »

1. In the results you will find the "Segment assignment". This first tab is the "FIMIX probabilities of segment membership". This is the native result of the algorithm which create a probabilistic segment assignment (i.e., each observation/case has a specific probability of belonging to each segment). The second tab contains the "Discrete segment assignment". Here we assign the observation to the group with the highest probability. You can copy this result into your dataset, create data groups in SmartPLS, and conduct an MGA or Permutation analysis to assess whether differences are significant. However, the results of this assignment might slightly differ from the FIMIX result because of the hard clustering. Observations with unclear segment membership (e.g., 51/49 probability) are now forced into one segment, which changes the results. This is also one of the downsides of FIMIX and why we usually recommend conducting a PLS-POS analysis on top of the FIMIX segmentation.
The FIMIX method itself has unfortunately no ability to assess whether differences are signifcicant.

2. You can also use one of these two results (probabilities of discrete assignment) to conduct your supplementary analysis to identify explanatory variables. For two segments, for example, I would recommend a logistic or probit regression with the assignment as DV and your additional explanatory variables as independent variables. But also decision trees (e.g., CHAID or C5) or simple cross-tables could help your understand which variables could describe the segments.

Please also read:
Hair, J. F., Sarstedt, M., Matthews, L., & Ringle, C. M. (2016). Identifying and Treating Unobserved Heterogeneity with FIMIX-PLS: Part I – Method, European Business Review, 28 (1), 63-76.
Sarstedt, M., Christian M. Ringle, and Hair, J. F. (2018). Treating Unobserved Heterogeneity in PLS-SEM: A Multi-method Approach. In H. Latan & R. Noonan (Eds.), Partial least squares structural equation modeling: Basic concepts, methodological issues and applications. New York: Springer.
Sarstedt, M., Radomir, L., Moisescu, O. I., & Ringle, C. M. (2021). Latent Class Analysis in PLS-SEM: A Review and Recommendations for Future Applications. Journal of Business Research, forthcoming.
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Post Reply