Indicators per Factor in Confirmatory Factor Analysis: More is not Always Better

By Jennifer Koran. Structural Equation Modeling: A Multidisciplinary Journal, 2020

69 citations recorded by the source corpus.

ABSTRACT Although some research in confirmatory factor analysis has suggested that more indicators per factor are generally better, studies have also documented that sample size requirements increase as model size increases. The present study used Monte Carlo simulation to investigate the effect of indicators per factor on sample size requirements. Results demonstrated a nonlinear association between the number of indicators per factor and the minimum required sample size while avoiding six important consequences for the analysis, such as bias in the model chi-square statistic. There is an upper limit for the desirable number of indicators per factor, and this upper limit depends on the number of factors and factor determinacy. The results showed clear patterns for the specific consequences that were most likely with too few or too many indicators per factor and inadequate sample size. Implications for further research are discussed.

Subfields: Computer Networks and Communications, Management Science and Operations Research, Organizational Behavior and Human Resource Management

View Paper: https://www.semanticscholar.org/paper/9edeb5f25e4fe36065638b91a43442e30a145a50