Level: Advanced
【Presenter】
Takenori Takahashi, Visiting Professor, Graduate School of Health Management, Keio University
[Announcement summary]
The main purpose of selective bilateral causal analysis in CS (Customer Satisfaction) survey is "to objectively clarify which target should be dealt with". For this reason, we focus on the causal items (services, product characteristics) that have a particularly strong influence on the result items (satisfaction evaluation), and actively exclude those that do not have an influence in the analysis process (The). We take the approach of vital few and the trivial many.).
However, it is valuable to grasp the bird's-eye view causal structure of CS by utilizing all the data collected in the CS survey after the completion of the selective bilateral causal analysis, in that advanced and useful knowledge can be obtained. As a useful method for this purpose, we propose a method based on SEM (Structural Equation Modeling) by reviving and utilizing what was excluded in the process of selective bilateral causal analysis.
Those that have an influence on the factors of the causal system (causal items and factors) that do not have a strong direct influence on the result system provide valuable information for understanding the overall causal structure. The start of path diagram creation for SEM after selective bilateral causal analysis is the processing of the structural model diagram of the final result of selective bilateral causal analysis. Those excluded in the analysis process are sequentially restored to this, and a bird's-eye view causal structure is constructed step by step. As is clear from the above, the essence of this method is a reversal causal analysis in which the causal structure obtained by selective bilateral causal analysis is the core, and the factors behind it are focused on and the figure is inverted. Although it is not SEM itself, it can be regarded as a quasi-SEM because it is essentially an approach similar to SEM.
* No video will be released in this announcement.
