Page 1 of 1

MICOM with more than two groups

Posted: Tue May 31, 2016 9:42 am
by humcasma
With SmartPLS 3 I can run the MICOM procedure to determine measurement invariance between two groups. In my case, I have 6 groups (6 countries), so I wonder if I could run MICOM for each pair of countries or if I need to run MICOM manually, including all groups in the permutation process (as it is done in the OTG approch to multigroup analysis). If I have to go for the latter approach, the number of permutations to perform will explode, right? I guess I could do as in the OTG approach randomly select and run, e.g., 5000 permutations only.

UPDATE: I have observed that sometimes (i.e., for certain groups) SmartPLS does not show the p-value for some LVs in the Step 3 results. Is this a bug? A problem with the data? (To clarify: 1 model, 3 groups. MICOM between group 1 and 2 shows all p-values; MICOM between group 1 and 3 does not show all p-values)

Any feedback will be extremely appreciated.

Cheers,
Nicolas