MGA unequal group sizes

Frequently asked questions about PLS path modeling.
Post Reply
Oemsi123
PLS Junior User
Posts: 2
Joined: Sun Sep 02, 2018 4:22 pm
Real name and title: Lea Schäffner

MGA unequal group sizes

Post by Oemsi123 » Sun Sep 16, 2018 10:27 am

Hello,

I run a MGA and found two significant differences in path coefficients that are quite interesting.
The problem is that the group sizes are small and not the same (N = 51 vs. N = 29).
I wonder now what exactly happens if I still run a MGA: Is there a chance that the results are wrong? Or that not every significant difference will be found?

Since the results are interesting I am thinking how I can report them anyway:
Would it be safe to report the path coefficients for each group without referring to the p-values?
Would it be acceptable to report the results and alert that the conditions were not great for running a MGA?
Or would you advise against reporting those results?

Thank you for your opinion!
Lea

jmbecker
SmartPLS Developer
Posts: 879
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: MGA unequal group sizes

Post by jmbecker » Sun Sep 16, 2018 3:54 pm

If you have such small sample sizes for the group results are certainly not very reliable. The biggest chance is that you are missing out important differences because of low statistical power. However, if your groups are not very representative for the general population for which they stand, you might also get biased (wrong) results. The second is often very hard to answer.
And of course, always when you are conducting a test where you are looking at the 95% confidence interval (5% error rate) you have a 5% chance of getting a wrong conclusion. That is the nature of such tests.
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

Post Reply