Can you please recommend a text or article citation for cross loading of items on factors? I have been told (courses) and seen (Nunnally 1994 cited, but no page number) that there should be at least a .2 difference if cross loading occurs. I have eyeballed Nunnally, but can not find that criterion.
I don't know of any useful rules regarding the size of cross-loadings. I would look at interpretability and significance. You might also want to look at the Cudeck and O'Dell paper that is referenced in the user's guide.
I think rules about sizes of loadings may be from a time when there were no standard errors available for estimated loadings - now they are available, also for EFA which is the topic discussed in Cudeck and O'Dell.
Substantively, of course you don't want large cross-loadings if they were not hypothesized, but there may be a good substantive interpretation of their existence and inclusion. They are not hurtful per se.
Mahdi posted on Wednesday, April 16, 2014 - 12:21 am
Hi What is the Cross-Loadings? and how do we compute them? Thanks a lot!
Say that you have 2 factors and expect the first 3 items to all load on the first factor. But EFA reveals that one of those 3 items also has a small but significant loading on the second factor. That is a cross-loading.