Abstract
Researchers often stray from recommendations provided by simulation studies when conducting confirmatory factor analysis (CFA), causing unwieldy applications of the analysis and diminished confidence in published results. We introduce three particularly important informal practices associated with (1) alternative interpretations of model fit, (2) the use of inadvisable combinations of fit indices, and (3) the failure to conduct effective model comparisons. We then review over 2000 CFAs in premier organizational science journals. Our results support that researchers widely engage in all three informal practices. To address this tension, we (1) formalize modern interpretations of model fit by providing percentile ranges of indices in published articles, such that researchers can make relative and continuous assessments of model fit. We (2) emphasize the importance of assessing multiple recommended fit indices together to provide complete depictions of model soundness. Lastly, we (3) demonstrate the necessity to perform appropriate model comparisons, including the assessment of more complex models.