Sorry if these end up being kind of naive questions, but I'm only starting to get into this type of data analysis technique:
-When I decide to remove a specific variable from my Rotated Component Matrix because of considerable cross-loading (therefore a complex variable), should I run the whole thing (PCA) again taking the said variable out from the start, or should I simply proceed with my original analysis only removing that variable's line from the Rotated Component Matrix?
-I've read that, ideally, the difference between the primary target loading of a variable and other secondary loadings should be greater 0.2 (lets call this the "0,2 rule" for easier understanding). Now my question here is how does this work in cases with positive/negative signs. For example: variable A loads primarily to factor 3 at .530, should the fact it also loads to factor 8 at -.356 concern me somehow? I know that if it were positive (.356) it certainly should. I also already know that any loading above the .3 range should be taken in consideration, so my question here is about the "0.2 rule" specifically.