Kennedy Balzen I am not very sure, as I haven't used Mplus, although I had performed and guided lot in these modification indices and other models, with that knowledge let me try to uncover your issues. Modification index (MI) value of 999.000 indicates a boundary condition, that is, the software has identified an issue with the model that prevents it from estimating the parameter in question. In your case, this can occur when there is either:

With the WLSMV estimator, this often points to an instance where the model cannot improve by adding or modifying the parameter. Since all other values (such as the Expected Parameter Change, or EPC) are 0, this suggests that no adjustment to the model will improve the fit for this parameter.

Given the MI is unusually high for one item, it may imply that the factor loading of this item on the latent factor could be constrained or problematic. This is why the reviewer might have been highlighted. Consider whether this item is redundant with respect to your latent construct, and you may want to explore model re-specification options, such as removing or modifying the item, if theoretically justifiable.

Be sure to document this anomaly in your report, as it may impact your conclusions regarding the model’s structure and item validity.

The MI value of 999.000 for the item with a low factor loading indicates that allowing this item to covary with other items could significantly improve the model fit, suggesting a substantial amount of unexplained variance in this item that is not captured by the current model (Kline, 2015). The zero values for the Expected Parameter Change (E.P.C.) and other parameters imply that the item is not effectively contributing to the latent factor, indicating it may not align well with the construct being measured. This situation warrants further investigation, such as reviewing the item's wording or conceptual relevance, to ensure that it is appropriately assessing the underlying construct.

Kline, R. B. (2015). Principles and Practice of Structural Equation Modeling (4th ed.). Guilford Press.

4o mini

Kline, R. B. (2015). Principles and Practice of Structural Equation Modeling (4th ed.). Guilford Press.

The MI value of 999.000 for the item with a low factor loading indicates that allowing this item to covary with other items could significantly improve the model fit, suggesting a substantial amount of unexplained variance in this item that is not captured by the current model (Kline, 2015). The zero values for the Expected Parameter Change (E.P.C.) and other parameters imply that the item is not effectively contributing to the latent factor, indicating it may not align well with the construct being measured. This situation warrants further investigation, such as reviewing the item's wording or conceptual relevance, to ensure that it is appropriately assessing the underlying construct.

Thanks so much, Anitha and Sasidharan. Interestingly enough, the item with this MI was not performing as a "problematic item" otherwise - the factor loading for the item was strong... I appreciate the references provided and suggestions!