Hi,

I am new to SEM and I have a question regarding individual-level residuals. I would like to analyse survey data using SEM in Stata. I want to include three latent variables in my analysis. Variable A is measured by 4 items in the questionnaire, B is measured by 3 items, and C is also measured by 3 items. The respective items will be the indicators of A, B, and C in my model. Before conducting SEM analysis, it is recommended to check that the individual-level residuals of the indicators are multivariate normally distributed. One way to do that is to use regression. Does using regression apply to a dataset like the one I have described? If yes, what type of regression needs to be conducted? If regression is not the correct way to get individual-level residuals in this scenario, which is the correct way? Thank you very much, Kyriaki