I am running a non-recursive model with clustered S.E. (type=complex, estimator=MLR), which has an interaction in one of the equations. And I am trying to retrieve the covariance matrix of the parameters via TECH3 to plot the conditional effects of the interation.
My problem is in the matching between TECH1 and TECH3. I've found in the user guide and in previous discussions here that TECH1 brings the list of parameters that are only numbered in TECH3.
However, my TECH1 does not have any numbering. Of course, I then assume that the numbering is the order by which parameters are listed in TECH1.
1) is such assumption correct?
2) If yes, the problem becomes: while TECH1 gives me matrices with 11 parameters, TECH3 gives results from 1 to 18. Other specifications Always do the same, i.e. give me more parameters in TECH3 than the listed in TECH1. How can I know, in this case, which are the 18 listed in TECH3? Maybe the first 11 are the 11 listed in TECH1, but in this case what are the extra 7?
Please, any suggestion of sources that may help me figure this out will be very much welcome.
As reference for future users I will mention here that relying on the order of printed parameters in TECH1, as I read people mentioning, can be misleading depending on the model specification.
And when using MplusAutomation in R, one has to be careful: parameter IDs from TECH1 are stored at the slot $parameterSpecification. Btw the process of getting parameters ID from tech1 and using in tech3 could be easily automatized as well.