Flexible cutoff values for fit indices in the evaluation of structural equation models

Researchers often struggle when applying ‘golden rules of thumb’ to evaluate structural equation models. this paper questions the notion of universal thresholds and calls for adjusted orientation points that account for sample size, factor loadings, the number of latent variables and indicators, as well as data (non-)normality. this research explores the need for flexible cutoffs and their accuracy in single- and two-index strategies. study 1 reveals that many indices are biased; thus, rigid cutoffs can become imprecise. flexible cutoff values are shown to compensate for the unique distorting patterns and prove to be particularly beneficial for moderate misspecification. study 2 sheds further light on this ‘gray’ area of misspecification and disentangles the different sources of misspecification. study 3 finally investigates the performance of flexible cutoffs for non-normal data. having substantiated higher performance for flexible reference values, this paper provides to managers an easy-to-use tool that facilitates the determination of adequate cutoffs.