The organization where I'm working at required me to study Factor Analysis to apply it to some of their research. They gave me different textbooks as study material and switching through both of them, I've noticed conflicting information.
One of them, by Hair, Black & Anderson, says that you need normally distributed data only if, later, you're going to apply some statistical test for the significance of factors that then assumes normally distributed data.
The other one, in my native language, tells me that you need normally distributed data to begin with, that is, factor analysis requires normal data just to start, and then offers SAS code to test for normality.
Now, I'm obviously no expert on Factor Analysis and now I'm confused: Do the data need to be normally distributed to begin with or not?