Fit indices (CFI, TLI, RMSEA, SRMR)

They show how well a model fits the data. CFI and TLI compare the model with a model without any relationships (higher is better; rule of thumb from .95). RMSEA and SRMR measure the misfit (lower is better; rule of thumb RMSEA up to .06 and SRMR up to .08, after Hu and Bentler 1999). The 90% interval of the RMSEA shows how precisely it is estimated.

Example: CFI = .97, TLI = .96, RMSEA = .04, SRMR = .04: good fit.

Caution: These are rules of thumb, not tests; with small models or few people they are unreliable. The χ² test is significant even for small misfit in large samples.