偏最小二乘路径模型中的分类变量：客户满意度研究的应用

raditional PLS path modeling-based customer satisfaction model could only handle continuous or Likert scale like variables. This study proposed an algorithm called categorical-variable partial least squares (CPLS) to handling categorical variables in PLS path modeling. Both categorical and numerical variables can be jointly analyzed in the same model and the proposed algorithm is suitable for both Mode A (reflective measurement model) and Mode B (formative measurement model). This algorithm was illustrated empirically in the context of a household appliance brand's customer satisfaction study. A complete process of using CPLS path modeling including model development, model assessment and the applications of the results were demonstrated. Moreover, the authors concluded with the limitations of their study and suggestions for future research.