The Performance of the SRMR, RMSEA, CFI, and TLI: An Examination of Sample Size, Path Size, and Degrees of Freedom

Keywords

structural equation modeling, growth model, sample size, degrees of freedom, simulation, science and math education

Abstract

The SRMR, RMSEA, CFI, and TLI are commonly used fit indexes reported when describing the fit of structural equation models (SEM) used in math and science education. A large number of the models tested in math and science education tend to be path models that study the interaction between various motivational, affective, contextual, and cognitive variables or latent growth curve models that examine change in students over time. The majority of these models tend to have small degrees of freedom and small sample sizes. Given the common use of these fit indexes, it is important to understand their performance when reported for relatively simple models.

(top)

(back to issue)