Outlier calculation in PLS-SEM

[sbax123]

Hi:
I am creating a new topic as it does not look like my specific query has been completely answered - please point me in the right direction if if already has :(

I am a PhD candidate with 650 samples/cases imported into SmartPLS3 and model drawn. I am following the steps in the Partial Least Squares Primer book and am up to data examination.

I have 4 missing values all up (in the demographic items)
Most of the data are responses to 7pt likert-like scales

I calculated Mahalanobis’ distance using SPSS Explore and it says that there are 78 outlier cases (I don't really want to get rid of 78 samples at this point).

I am just confused whether I actually needed to do this, as mentioned in a previous post. Could some one please give a direct answer for this?

Thank you,
Sam

[cringle]

Sure, you can do this. But the question is if the data really is an outlier or an extreme response pattern. This question never is easy to answer. You may want to look for certain response pattern (e.g., "straight lining") or misstakes in the data (e.g., a 77 instead of a 7 on a 1 to 7 scale).

Also FIMIX-PLS can help you to identify extreme data. If you run FIMIX-PLS say for 10 or 20 segments, you will obtain micro segments with extreme observations. They might be worth to be inspected in more detail.

Best
CR