Partial least squares in supply chain management: an example study on sustainable supply chain management practices

In recent years some scholars have raised concerns over the use of the PLS technique. Criticisms of traditional PLS methods are due to them being composite-based, not factor-based. That is, in traditional PLS methods, latent variables are estimated as weighted aggregations of indicators without the inclusion of measurement errors. The measurement errors usually serve as extra indicators that often complement the actual indicators; together, the actual indicators and measurement errors constitute factors. Without considering measurement errors, the use of composites instead of factors leads to some known sources of bias. The path coefficients tend to weaken with respect to their corresponding true values. Thus, recent methodological developments building upon traditional PLS techniques have helped to bridge the gap between factor-based and composite-based structural equation modeling (SEM) techniques. To substantiate our claim we have used data gathered using a structured questionnaire to test a theoretical model which we have grounded in resource-based view (RBV) and institutional theory (IT) under the mediating effect of top management commitment. We used WarpPLS 5.0, which is a popular PLS technique that has been recently used for path-analytical models. The results obtained via PLS-SEM analysis were examined and compared with existing literature. We hope our attempt will help management scholars who want to use PLS-SEM and embrace empirical study to build theory to explain complex operations and supply chain management phenomena.