Abandon All Thumbs Ye Who Model: CFA fitting for the new century
Confirmatory factor analysis (CFA) is one of the most central analytic techniques in social sciences. However, model fit assessment still heavily relies on arbitrary cutoffs (e.g., RMSEA ≤ .05, CFI ≥ .95), which lack universal validity. Misinterpreting these thresholds contributes to the accumulation of misspecified models in the literature. In our talk, we introduce an up-to-date CFA fit assessment framework based on conceptual understanding, local fit evaluation, and dynamic benchmarks. Our tutorial comes with all the resources needed to bring CFA-fitting into the new century: an R script, empirical examples, and a decision tree. By abandoning inflexible cutoffs and embracing a more nuanced model fit evaluation, we can shrink the number of misspecified models in the literature and, consequently, refine our theories.