Evaluation of measurment and structural model of the reflective model constructs in PLS – SEM

Abstract

This paper reports the procedure ofthe assessment of measurement model and the structural model of the reflective constructs by using newly origin second generation multivariate statistical technique of PLS – SEM(Partial Least Squares - Structural Equation Model). As a conceptual paper, it highlighted the differences between the concepts of covariance based SEM and variance based SEM, measurement model and structural model focusing reflective models. The general objective of the study is to investigate the sequential steps and prerequisites of convergence of any PLS - SEM based model. As a newly origin statistical tool (by using PLS 3 version analysis) the procedure and required steps to follow is ratherscant hence,this paper may provide appropriate and accurate guidance for Sri Lankan potential researchers. Study was based on a conventional review and analysis based on the extant literature froma series of texts which are obtain reviewing different data bases. As per the discussion basically, two main criteria called reliability and validity have to be achieved in measurement model before evaluatingthe structural model. Internal reliability and composite reliability scales were commonly employed to asses construct reliability of the intended constructs. However, convergent validity achieved through Average Variance Extracted and factor loadings. Discriminant validity can be evaluated by assessing the cross loadings among constructs, Fornel-Larcker criterion, and Heterotrait- Monotrait Ratio of correlation (HTMT). After satisfying prerequisites of measurement model analysis have to proceed the evaluation of thestructural model. In order to evaluate the structural model basically have to follow five stepsas assessing astructural model for collinearity issue ,assess the path co efficient,assess the level of R2 ,assess the effect size f2 ,assess the predictive relevance Q2. All the threshold values against to each and every criterion were clearly represented under the conclusion to have comprehensive understand about the evaluation of measurement and structural model.