Is it possible to detect outliers with missing data?

[atanasj]

I am trying to diagnose some issues with a lavaanmodel, however I age getting the error:

Error in forward.search(phd_fin_wide, riclpm_mod) : 
  All routines require complete datasets (no NA's) so that the search
             gives meaningful results.

I can run the model in lavaan using fiml… am I doing something wrong?

[philchalmers]

Could you provide a reproducible example of this issue? It's possible that some methods can work well with missing data, such likelihood-based criteria, but others likely will no work well. In any event, some simple example to reproduce the problem would be appreciated.