I was trying to make sense of a simple mediation model that I tried to fit onto data. I investigated the relation with path analysis in R with the lavaan package in which I specified the following paths:
y<-b*M+cp*X
M<- a*X
for the effects:
ab := a*b
total:= a*b+cp
So the output gives me a chi square of 0.000 for the model. I thought this is because the model is just-identified/saturated. At first, I thought that this was no problem since I was going to compare it to more complex models via CFI/TLI/AIC...but with every further google search, I got more unsure whether it is okay to consider this model hence it fits the data the best (according to regular SEM theory).
When I also looked more closely into the output, I got confused of the number of model parameters, or more precisely, the amount of df because the output says that there are 0 dfs even though acc. to the formula to determine the amount of dfs there should be 1 left (df=#observations (=6)- #estimated parameters (=5)). Every help would be greatly appreciated!