On tests and indices for evaluating structural models

Eight recommendations are given for the improved reporting of research based on structural equation modeling. These recommendations differ substantially from those offered by Prof. Barrett in this issue, especially with regard to the virtues and limitations of current statistical methods.