Skewed continuous outcome in mediation (SEM)

[Anna Warren]

Hello,

I am conducting a mediation analysis using SEM. My continuous outcome is skewed. I wondered whether this is a problem and if so how I should deal with this. I have read in some places that it does not matter if the outcome is normally distributed as long as the residuals are normally distributed? Is this correct?

Thank you,
Anna

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

SEM usually assumes the variables (not just the errors as in most econometrics) are normal. This is clear in most texts on the technique. How much of a problem this is I think has been debated. You can either transform variables to be closer to normal, or use gsem.