Grant Open

Extending Dynamic Fit Index Cutoffs for Latent Variable Models

NCER

Program: Statistical and Research Methodology in Education

Program topic(s): Core

Award amount: $557,555

Principal investigator: Daniel McNeish

Awardee:

Arizona State University

Year: 2022

Project type:

Methodological Innovation

Award number: R305D220003

Purpose

The goal of this proposed project is to provide empirical researchers with more generalizable, accessible, and accurate guidelines for assessing model fit in latent variable models. Traditional cutoffs have multiple known shortcomings, including the tendency to favor models with low construct reliability over those with high construct reliability. To address some of the drawbacks of traditional fit indices, the research team will expand the scope, testing, and software availability of a relatively new approach called the dynamic fit index (DFI). The DFI is a simulation-based technique that adapts to the context of a latent model to determine fit cutoffs which are specific to the model being tested.

Project Activities

Current software can handle only continuous indicators. The primary DFI expansion in this project is the development of software that allows for estimation of latent variable models that have categorical factor indicators; The researchers will also adapt DFI for use in testing measurement invariance across two groups. After these approaches are derived and programmed into software, the team will use Monte Carlo simulations to investigate the performance of DFI relative to more traditional fit indices.

People and institutions involved

IES program contact(s)

Charles Laurin

Education Research Analyst

NCER

Products and publications

Products: The researchers plan to create both a Shiny app for applied researchers and an adaptation of the R package lavaan for use by methodological researchers who are interested in exploring additional uses of DFI.

Publications:

McNeish, D. (2023a). Dynamic fit index cutoffs for categorical factor analysis with Likert-type, ordinal, or binary responses. American Psychologist, 78(9), 1061-1075.

McNeish, D. (2023b). Generalizability of Dynamic Fit Index, Equivalence Testing, and Hu & Bentler Cutoffs for Evaluating Fit in Factor Analysis. Multivariate Behavioral Research, 58(1), 195-219.

Wolf, M. G., & McNeish, D. (2023). dynamic?: An R Package for Deriving Dynamic Fit Index Cutoffs for Factor Analysis. Multivariate Behavioral Research, 58(1), 189-194.

Publicly available data:

McNeish, D. (2022, June 21). Traditional vs. Modern Methods for factor analysis model fit. Retrieved from osf.io/x758r