Spearman Rank Correlation PCA for Mixed Scale Indicator in Structural Equation Modeling

Structural Equation Modeling (SEM) is a statistical modeling technique that integrates measurement models and structural models simultaneously. In the SEM measurement model, not all latent variables are metric, they can be mixed scales, namely metric and non-metric which have not been widely studied. This study aims to apply the Spearman Rank Correlation Principal Component Analysis (PCA) to handle mixed-scale indicator data in a mixed measurement model (formative and reflective). This method is evaluated on a case study of fertilizer repurchase decisions, resulting in a total determination coefficient of 80%. This shows the flexibility of SEM in handling the complexity of mixed-scale data without sacrificing estimation accuracy. The results showed that the Spearman Rank Correlation PCA was able to store 78.62% of the diversity of data from mixed-scale indicator variables, namely Farmer Demographics (X2). In addition, the results showed that Customer Satisfaction (X1) significantly influenced Repurchase Decisions (Y2) but did not directly affect Customer Engagement (Y1). Farmer Demographics (X2) significantly influences Customer Engagement (Y1) and Repurchase Decisions (Y2), and Customer Engagement has a significant effect on Repurchase Decisions (Y2).