Convenient function for the test of small differences in fit as proposed by MacCallum et al. (2006).
This function is based on the R-Script by Dr. Timo Gnambs: https://timo.gnambs.at/research/test-of-small-difference-in-fit-sem
This function depends on "lavaan" and "semTools", both of which you'll probably already have loaded at this point.
The function only needs two arguments to be specified: fit1
and fit2
, which are lavaan fit-objects.
groups
is set to 1 by default and can be changed if your fits have more than one group.
alpha
, rmseaa
and rmseab
also have default values. I recommend against changing these, but you can, if you wish to.
robust
is also by default set to be TRUE
, but will only work if the fitted lavaan objects contain the information for robust variants of chi-squared. This can be achieved by setting estimator = 'MLR'
when fitting the object.
Feel free to contact me for questions or suggestions.