Frequently asked questions about PLS path modeling.
Post Reply
PLS Junior User
Posts: 1
Joined: Fri Mar 17, 2017 2:32 pm
Real name and title: Ms. Saadia Shabnam


Post by Saadia » Fri Mar 17, 2017 2:59 pm

Good day.
I am using SmartPLS 3. Before running PLS-MGA, ran permutation test and got MICOM output (Henseler et al, 2016). From stage 2, when result is showing 'no' compositional invarience, how to treat it and proceed for MGA ?
Can any indicator level change (omitting certain indicator from constructs which is showing no compositional invarience)?
Quoting from Henseler et al, 2016-
" lack of compositional invariance implies that the scores obtained through group-specific model estimations differ from the scores resulting from the pooled data (no measurement invariance established). Researchers should only analyze and interpret the group-specific model estimations separately."

Any clarification on this? any other resource please?

Thanks in anticipation.

User avatar
SmartPLS Developer
Posts: 805
Joined: Tue Sep 20, 2005 9:13 am
Real name and title: Prof. Dr. Christian M. Ringle
Location: Hamburg (Germany)


Post by cringle » Fri Mar 17, 2017 8:38 pm

Are you sure that you did not establish compositional invariance? This is not the case, when the original correlation is below the 5% lower bound. Otherwise compositional invariance has not been established.

When compositional invariance has not been established, you cannot tell if the group-specific differences are significant (e.g., by permutation or PLS-MGA). But you can qualitative discuss the differences. Alternatively you can try solving the problem by analyzing a reduced model that does not include the problematic constructs.


PLS Junior User
Posts: 2
Joined: Fri Jan 19, 2018 8:54 am
Real name and title: Lintje Sie


Post by sling » Sun Jan 21, 2018 8:50 pm

Hi Prof Ringle,
Just to follow above conversation, my result of MICOM analysis showed compositional invariance (step 2 established) but step 3 did not achieve (unequal mean values and variances). Does it mean I should only analyze and interpret the group-specific model estimations separately? For your immediate advice, please. Thanks.

Best regards,

SmartPLS Developer
Posts: 971
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker


Post by jmbecker » Mon Jan 22, 2018 8:37 am

If you have established compositional invariance (step 2) you can use the MGA or permutation approach to assess the significance of the difference. However, you should not combine the groups and assess the overall (combined) sample.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

PLS Junior User
Posts: 8
Joined: Wed Dec 19, 2018 2:23 pm
Real name and title: Maria Lunde


Post by MariaLunde » Fri Dec 21, 2018 11:18 am


What do you mean with "Combine the groups and assess the whole sample"? For example, can we run multigroup analysis for one variable with four categories?

I am really looking forward to your reply:)

kind regards,

Post Reply