This paper addresses issues related to goodness of fit in structural equation models (SEMs) with unobservable variables and measurement error. It challenges the validity of Bagozzi's criticisms of the Fornell-Larcker Testing System, arguing that the criteria he proposes are not relevant or reliable for assessing model fit. The paper emphasizes that the Fornell-Larcker system is internally consistent and aligns with the rules of correspondence for relating data to abstract variables. The paper discusses the algebraic and statistical properties of factor analytic structural models. It shows that the chi-square statistic does not evaluate the strength of relationships but rather the structure of underlying relationships. Structural consistency, determined by the divergence of observed correlation coefficients, is the key factor in model fit. Bagozzi's claims about the "anomalous" findings of simulations are shown to be based on misunderstandings of the role of correlation matrices. The paper also addresses the issues of convergence and differentiation in SEMs. It argues that these criteria are not relevant to the properties of the chi-square statistic and that the evaluation of structural equation models should not rely solely on raw correlation coefficients. Instead, the paper advocates for a more comprehensive approach that considers both types of errors (systematic and random) in the model. The paper discusses the importance of measurement error in SEMs and how it affects the estimation of structural parameters. It argues that the ability to remove measurement error from theory testing procedures is a key methodological contribution of SEMs. The paper also highlights the importance of validity assessments in SEMs, particularly in determining the extent of measurement error. The paper concludes that the Fornell-Larcker Testing System is a valid and reliable method for evaluating SEMs with unobservable variables and measurement error. It also emphasizes the importance of considering the research objective in determining the choice of interpretative statistics. The paper argues that the proposed testing system is not limited to causal modeling and that empirical relationships can be examined at the level of observations. The paper also discusses the importance of generalizations in SEMs and the challenges associated with providing a single measure of overall explanatory power in complex multivariate relationships. Finally, the paper highlights the importance of measures of explanatory power in full information maximum likelihood estimation of structural equations.This paper addresses issues related to goodness of fit in structural equation models (SEMs) with unobservable variables and measurement error. It challenges the validity of Bagozzi's criticisms of the Fornell-Larcker Testing System, arguing that the criteria he proposes are not relevant or reliable for assessing model fit. The paper emphasizes that the Fornell-Larcker system is internally consistent and aligns with the rules of correspondence for relating data to abstract variables. The paper discusses the algebraic and statistical properties of factor analytic structural models. It shows that the chi-square statistic does not evaluate the strength of relationships but rather the structure of underlying relationships. Structural consistency, determined by the divergence of observed correlation coefficients, is the key factor in model fit. Bagozzi's claims about the "anomalous" findings of simulations are shown to be based on misunderstandings of the role of correlation matrices. The paper also addresses the issues of convergence and differentiation in SEMs. It argues that these criteria are not relevant to the properties of the chi-square statistic and that the evaluation of structural equation models should not rely solely on raw correlation coefficients. Instead, the paper advocates for a more comprehensive approach that considers both types of errors (systematic and random) in the model. The paper discusses the importance of measurement error in SEMs and how it affects the estimation of structural parameters. It argues that the ability to remove measurement error from theory testing procedures is a key methodological contribution of SEMs. The paper also highlights the importance of validity assessments in SEMs, particularly in determining the extent of measurement error. The paper concludes that the Fornell-Larcker Testing System is a valid and reliable method for evaluating SEMs with unobservable variables and measurement error. It also emphasizes the importance of considering the research objective in determining the choice of interpretative statistics. The paper argues that the proposed testing system is not limited to causal modeling and that empirical relationships can be examined at the level of observations. The paper also discusses the importance of generalizations in SEMs and the challenges associated with providing a single measure of overall explanatory power in complex multivariate relationships. Finally, the paper highlights the importance of measures of explanatory power in full information maximum likelihood estimation of structural equations.