Though outliers are part of the work as they help to enrich your interpretations. But Factor Analysis experts remove these outliers as they can destruct the results of the study. When an item from the communalities table has less than 0.4 they are removed and other items are extracted. This shows that the removed items do not load well in the initial extraction and might not load well on any component if extracted for use in the study.

Check this..

Dear Chris,

Some statistical tests are sensitive to outliers. So they will violate the assumptions of your statistical analysis.

But, If you want to keep outliers to capture the full variability out of your data, you can take some points into account, as follows:

1. Nonparametric tests are robust to outliers.

2. You can transforming your data or using a robust regression analysis.

3. bootstrapping techniques.

Thanks for your answers!

How would you proceed if new outliers are detected after you already removed some? Remove them again? or just for the first time!

Thanks!

Chris Cl, I can see why you are facing difficulties. There are a number of things to consider. I found it interesting that Goodwin and Leech (2006) stated that outliers can affect the whole family of statistics related to correlation, including regression, factor analysis, and structural equation modelling.

Here's the reference for them:

Goodwin, L. D., and Leech, N. L. (2006). Understanding correlation: Factors that affect the size of r. Journal of Experimental Education, 74(3), 251–266. https://doi.org/10.3200/JEXE.74.3.249-266

It might be worth running your analyses with the outliers in, then with the outliers having been dealt with e.g., by removing them or by winsorising them - and seeing whether you get different results. If you get different results, you'd need to think carefully about which of those results you want to trust more - and that shouldn't be determined by what your personal wishes are.

All the best with your research.