This study proposes a new method in PLS SEM segmentation, namely PLS SEM Kernel K-Means Clustering (PLS SEM KKC). Segmentation is conducted to overcome one of the main limitations of PLS SEM modeling: unobserved heterogeneity. Previous studies on segmentation in PLS SEM have been performed using a linear clustering method. Segmentation is carried out based on the residual values of measurement and structural models from global PLS SEM, where the characteristics of the residuals are non-linear; thus, the segmentation process requires non-linear segmentation. This study's primary contribution is integrating kernel-based clustering into PLS SEM segmentation. The method effectively addresses unobserved heterogeneity by capturing non-linear residual patterns, leading to more accurate models. The empirical results show that the PLS SEM KKC method significantly improves model accuracy, with R² increasing from 51.1 % (global model) to 93.9 % (k = 2) and 97.5 % (k = 3) in segmented clusters. The increase in local R² confirms overcoming unobserved heterogeneity by grouping observations with similar patterns into homogeneous segments, improving model accuracy. • This study recommends a new method in PLS SEM segmentation. • PLS SEM KKC effectively captures non-linear residual patterns to address unobserved heterogeneity and improve model accuracy.
Astuti et al. (Fri,) studied this question.