ANOVA in addition to Multi-Group Analysis?

[emmabr]

Dear all,

I am currently assessing the results of my study through SmartPLS. My goal is to compare three groups. Therefore, I have adjusted the error level according to the Bonferroni adjustment, conducted MICOM and ran the permutation-based MGA to compare the path coefficients pair-wise. The results show no significant differences in path coefficients.

However, apart from path coefficients, I would also be interested in the "absolute" differences of constructs between groups, especially the DV (acceptance). Would it be appropriate to use a one-way ANOVA for this? Reading from the book Advanced Issues in Partial Least Squares Structural Equation Modeling, it was mentioned conducting an ANOVA is not meaningful. However, the OTG was given as a remedy to this, which I now read is refuted as well.

Therefore, for the purpose of comparing absolute evaluations between the groups for further knowledge of the results, would it be appropriate to conduct an ANOVA next to MGA?

Thank you in advance!

[jmbecker]

Doing an ANOVA on the latent variable scores would be possible, but it would not be in line with the non-parametric nature of PLS and probably not giving you the best result.

However, you get information about mean differences from the permutation MGA invariance testing: Step 3a (mean).
This is basically a pair-wise test of mean differences that you are looking for.