ModIndices include existing parameters - why?

[Lukas Wallrich]

Dear all,

I am quite new to Sem and lavaan and struggling to make sense of some modification indices.

My model includes:

ContValence =~ WhExpBlVAL

DivVALUE =~ DivValINSTR

 ContValence =~                                                        

    WhExpBlVAL        1.000                               1.705    0.948

   DivVALUE =~                                                           

    DivValINSTR       1.000                               0.965    0.920

(I included these single predictor latent vars to deal with measurement error/reliability by fixing the residual variance of the observed variable).

The top-2 modindices, however, are:

17       DivVALUE =~ DivValINSTR 217551.989  1590.777  1535.867  1464.288  1464.288

14    ContValence =~  WhExpBlVAL 110771.910 -1909.302 -3256.312 -1809.418 -1809.418

As I understand them, this suggests that I should consider adding those parameters - but they are there already? What does that mean then?

Relatedly, what do the LatentVar ~1 modindices mean? It's not quite clear to me what freeing them would mean, and if I try to do it in the code for more than 1 variable, the model is no longer identified.

Apologies if these questions are very basic and have been answered before, but so far I haven't found a good explanation online.

Many thanks and all the best,

Lukas

[Lukas Wallrich]

I just updated lavaan and rerun the analysis - now the list of modindices is completely different and makes a lot more sense to me.  So unless you tell me otherwise, I will assume that was a bug.

Thanks,

Lukas