Hi all,
I have two groups, both groups violated multivariate normality, so I run B-S bootstrapping for each CFA. When it comes to test measurement invariance, should I run bootstrapping? I didn't see output for skewness, kurtosis, multivariate kurtosis, and c.r. though.
In this case, how would it affect Δχ2? Am I still able to use this indicator? I also use other indicators such as ΔRMSEA and ΔCFI.
TIA
I can perhaps help but I don't understand your abbreviations CFA, B-S, c.r., DeltaRMSEA, DeltaCFI, DeltaChi2
Yuling Hu When it comes to test measurement invariance, should I run bootstrapping?
- Yes, since your data violate multivariate normality you should continue using BS bootstrapping when checking measurement invariance (the BS adjusts the χ2 making it more reliable for non-normal data) since Δχ2 assumes multivariate normality. You could also try a robust test instead, like a corrected Satorra-Bentler test. But again, ΔRMSEA and ΔCFI are generally accepted as invariance indicators and are not that prone to issues with non-normality
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science
- Reasoning