Validation of the BFI-10-R: A New BFI Scale with Strong Structural and Construct Validity

Bryant M. Stone,Emily M. Bartholomay,Amanda B. Chamberlain
DOI: https://doi.org/10.1007/s10862-022-09978-4
2022-06-22
Journal of Psychopathology and Behavioral Assessment
Abstract:The BFI-44 and BFI-10 are widely used, open-source measures of the Big Five Model of personality. Exploratory factor analyses and exploratory structural equation modeling produce models with non-negligible cross-loadings, creating unclear factor structures and confounded validity evidence. Further, no studies have independently confirmed the structural validity of the scale using confirmatory factor analyses. In Study 1, with a sample of community members ( n = 345), we demonstrated that the BFI-44 and BFI-10 fail to demonstrate acceptable data-model fit and contain an excessive number of correlational residual statistics (i.e., BFI-44: 36%; BFI-10: 42%). In Study 2, with a sample of undergraduate students ( n = 110), we created a version of the BFI-10 with strong structural and construct validity, the BFI-10-R. Using an EFA, we selected the top two items from the BFI-44 for each Big Five trait with the strongest loadings onto their factor and with the weakest cross-loadings. We then confirmed the new 10-item factor structure using a confirmatory factor analysis and found a model with strong global and local fit in both samples. This new model had few significant correlational residual statistics, strong convergent and divergent validity, and strong internal consistency. Researchers may consider using the BFI-10-R in place of the BFI-10 due to its preferable psychometric properties.
psychology, clinical
What problem does this paper attempt to address?