EFA Question - What to do with variables with low loading values across factors?

[matt starcke]

Hi all,

I'm conducting an exploratory factor analysis using 15 items from a survey. Running the

factor q1-q15, pf

command and looking at the subsequent scree plot suggests 3 factors are appropriate. After rotation, 13 of the 15 variables load onto these three factors at >.3 (the highest abs loading values for the remaining 2 variables are 0.2467 and 0.1866, respectively). My question, then, is how to handle these 2 variables? They are somewhat distinct from the other included variables, so it may make sense to remove them from the EFA process and include them as stand alone items in any final analyses (e.g., future logit models). That said, I'm relatively new to EFA, so I may be missing something. Appreciate any thoughts or suggestions you might have.

Thanks all.