How data analysis can dominate interpretations of dominant general factors

Wiernik, B. M., Wilmot, M. P., & Kostal. J. W.
Industrial and Organizational Psychology, 8(3), 438–445
(2015) http://doi.org/10/895

Dominant general factors (DGFs; where one latent contributes a majority of the variance across a set of measures) are a pervasive phenomenon in organizational research. Failure to properly model these factors in data analysis can have dramatic impacts on observed results, leading to erroneous substantive conclusions. In this paper, we demonstrate the shortcomings of several traditional research methods in the presence of DGFs using data on job satisfaction and job performance. We show that bifactor models are the only analytic method that accurately identifies the contributions of general and specific satisfaction factors to job performance.