Hi,
Looking for advice about path analysis in Mplus.
I have produced a model, reported the direct and indirect relationships, and have model fit statistics. However I am struggling to make sense of them and think I may have made an error.
The model fit results are as follows:
χ2 = < 0.001, p < 0.01; CFI = 1; TLI = 1; SRMR =
Your model is saturated (just identified). Therefore, the chi-square test has zero degrees of freedom and a test statistic of zero indicating perfect fit. The p value is incorrectly given as 0.000 by this Mplus version (It should be printed as 1.000.)
The reason you are getting this result as well as the other "perfect" fit statistics is because a saturated model by definition is not testable. It trivially fits the observed data perfectly because it uses as many parameters as there are pieces of information in the data (i.e., all possible direct paths are estimated in your model; no path is implicitly or explicitly fixed at zero). Therefore, the fit statistics are not meaningful for the model.
Nonetheless, the path coefficients and the corresponding direct, indirect, and total effects are interpretable and meaningful even for a saturated model. This is also true for their associated standard errors and z-tests of significance. So you can focus your interpretation on those rather than global model fit statistics.
Adam, you can ask help here and you will be assisted : https://stats4edu.com/services/structural-equation-modeling/
- Badges
- Science topic
- Similar topics
- Indexes