The purpose of this study is to provide an overview of emerging prediction assessment tools for composite-based PLS-SEM, particularly proposed out-of-sample prediction methodologies.
A review of recently developed out-of-sample prediction assessment tools for composite-based PLS-SEM that will expand the skills of researchers and inform them on new methodologies for improving evaluation of theoretical models. Recently developed and proposed cross-validation approaches for model comparisons and benchmarking are reviewed and evaluated.
The results summarize next-generation prediction metrics that will substantially improve researchers' ability to assess and report the extent to which their theoretical models provide meaningful predictions. Improved prediction assessment metrics are essential to justify (practical) implications and recommendations developed on the basis of theoretical model estimation results.
The paper provides an overview of recently developed and proposed out-of-sample prediction metrics for composite-based PLS-SEM that will enhance the ability of researchers to demonstrate generalization of their findings from sample data to the population.
