sam() with missing data

[Kyle Cox]

Hi all,

I'm trying to (1) find any documentation on how missing data are handled by the sam() function in lavaan beyond what is noted in description of the LSAM approach in recent publications and (2) confirm these methods are implemented in the current version of lavaan. I am mapping out estimator performance with SEMs that include latent interactions and quadratic terms when missing data are present but initial performance was concerning enough I am reviewing my simulation study methods but also wanted to check on issues with implementation of LSAM in the sam() specifically with handling missing data.

Thanks,

Kyle Cox  

[Felipe Vieira]

Hi, 

I am currently investigating this exact topic, as there appears to be a bug in the code. There is an issue open on GitHub about this as well. Theoretically, under specific missingness assumptions, it should work. 

Eventually, I will talk to Yves about this and come back with a more informative answer.

Best, 

Felipe.