Measuring Parameter Uncertainty by Identifying Fungible Estimates in SEM

Abstract

Researchers require methods for evaluating whether statistical results are credible and thus, worthy of interpretation. An examination of fungible parameters estimates is a method in which the veracity of inferences can be strengthened or weakened by quantifying the level of support for individual parameter estimates as measured by the likelihood function. This new approach aims to overcome some of the limitations of previous methods and is based on a simplified computational and conceptual understanding of a fungible parameter estimate analysis. Additionally, this computational approach has been implemented in the R package psindex. This package provides a user-friendly, and free method for statisticians and applied researchers to investigate parameter stability in their own models using an analysis of fungible parameter estimates. Finally, an applied example illustrates fungible parameter estimate analysis in practice using the psindex package. General guidelines for interpretation are provided.