Abandon All Thumbs Ye Who Model: An Up-to-Date Tutorial on Fitting CFA Models

Abstract

Confirmatory factor analysis (CFA) is the bread and butter of many psychology and social science researchers. To evaluate how the data aligns with the instruments’ hypothesized structure, researchers often rely on the so-called rules of thumb, model fit cutoffs, such as RMSEA ≤ .05 or CFI ≥ .95, whose validity has been widely criticized. The cutoffs have been derived from experience or based on simulation studies. However, it is impossible to use universal cutoffs for models and data with characteristics that differ from those used in prior experience or simulations. In this paper, we offer an approach to model fit assessment grounded in conceptual understanding and situational character of the fit measures. We hope that this tutorial, together with an R script with illustrative examples and a model-fitting decision tree, may inspire fellow researchers to conduct more in-depth fit assessments and prevent the accumulation of bad-fitting models in the literature.