what is the critical value for Satorra-Bentler Chi-Square Difference Test.Also are Mplus critical values z or t. I read some papers reporting it as t but i see z in the software output. which is which?
It depends on the two models that are being compared. You should look up in a chi-square table the critical value for the difference in degrees of freedom or the difference in the number of parameters for the two nested models.
The ratio of the parameter estimate to its standard error is a z-test in large samples.
Genta posted on Monday, September 23, 2013 - 7:03 pm
Hi, I am running a two-level random effects model and I am trying to compare the equivalency of parameter estimates across genders. I am using the Satorra-Bentler chi-square test (formula 1 in Webnote 12) to compare nested models with gender-specific vs. gender-equivalent estimates. The test produces negative values. I followed this note's recommendations for how to estimate a strictly positive test statistics (formula 5, example 3). I am having problems with model convergence. Any suggestions/feedback would be greatly appreciated! Thank you!