Selecting items for Big Five questionnaires: At what sample size do factor loadings stabilize?

Researchers often use exploratory factor analysis (EFA) to develop and refine questionnaires assessing the Big Five personality traits. We use sequential sampling and bootstrapping to determine the number of participants needed to yield stable loading patterns for the Big Five Inventory (BFI) and the International Personality Item Pool Big Five measure (IPIP). Overall 21,350 participants (BFI = 10,285; IPIP = 11,065) participated. In two studies primary factor loadings are highly variable in smaller samples (n < 500) and some primary loadings are not stable with 10,000 participants. Most studies will not have adequate sample size to yield stable loading patterns for Big Five measures such as the BFI and IPIP. Researchers should assess and report the variability of loading patterns.