Frequently asked questions about PLS path modeling.
Post Reply
PLS Junior User
Posts: 8
Joined: Tue Apr 14, 2015 8:43 am
Real name and title: nuria recuero virto. PhD


Post by NRECUERO » Fri Jan 19, 2018 12:33 pm

Good afternoon,

My doubt is regarding "Results of invariance measurement testing using permutation".
I have a PLS-MGA model, and I have just one indicator of Partial measurement invariance established that does not suit
Compositional invariance (Correlation = 1)
C=1 5% quantile
0.996 0.997

As we can observe, for this construct theoretically partial measurement invariance would not be established, which is rather interesting because it just a 0.001 below the requirements. Besides, would this indicator of 0.996 mean it is not reaching Configural invariance (Same algorithms for both groups)?

Thanks for your help,

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 » Tue Jan 23, 2018 11:39 am


(1) You need to compare the original correlation value with the lower bound of the confidence interval.

(2) The the original result is 0.996 and would be lower than 0.997 you would reject the null hypothesis of equal correlations on the 5% level. How about running more permutations (e.g. 5000)? You may also consider using a different significance level.

Best regards

Post Reply