I have an interesting question. I recently got a review back on a paper where we tested the moderating effect of GRIT on Person-Environment Fit, Happiness, and Psychological wellbeing via SEM. We adopted a competing measurement modelling strategy to determine the best model for our data. We included all our factors in each measurement model.

The reviewer requested that we remove GRIT as a factor from the competing measurement model estimation, and only include it in the final structural model (for estimating the moderating effect).

This seems a bit odd, and I dont really understand the reviewer's reasoning behind it. So my question is, is there a reason for such (couldnt find anything in the standard literature), and then what would the justification be if we exclude it from the measurement model estimations

(Because including or excluding Grit in the measurement models will obviously increase / decrease model fit)

Thanks for your help!