SEM is a modelling tool, and not a tool for “descriptive” analysis. It fits models to data. These models require testing in order to determine the fit of a model to the data. If an investigator refuses to use or acknowledge the results from the one statistical test available to assess model fit, then something other than some ad hoc scaled discrepancy index will be required to justify the worth of the model. I have provided the logic of how to approach assessing model acceptability if a sample