Greetings,

I'm running Stata 15.1 on OSX and working with cross-sectional survey data. I'm trying to test a mediation model with a binary outcome and continuous predictors. I'm thus using GSEM, which allows for non-continuous DVs. However, after exhaustively browsing this forum, I get the sense that one is very limited when it comes to assessing GOF when using GSEM. Specifically, unless I'm missing something, I appear to be limited to examining the information criterion via 'estat ic' and the path coefficients/effects via nlcom. Baseline comparison stats (CFI & TLI), the likelihood ratio stats (chi2_ms etc.), and modification indices are inaccessible with GSEM, which makes it rather difficult to determine whether my model fits the data well or needs improvement. I'm left wondering whether I'm better off treating my DV as continuous and sticking with SEM. Does anybody have any advice on how I should proceed? Thanks in advance!