Answers are grounded in this paper and cite their sources.
This article demonstrates that excluding greater cross-loadings may paradoxically improve model fit, indicating a new perspective on model misspecification.
In structural equation modeling or factor analysis, excluding nonzero cross-loadings is generally regarded as a form of model misspecification. It is also commonly believed that excluding a greater number or larger size of cross-loadings leads to a more severe model misspecification, thereby resulting in a poorer model-fit. However, a special form of cross-loadings in the population can be perfectly accounted for by the structural model without causing any model misspecification. By exploring this form, this article shows a paradoxical phenomenon in which a model that excludes greater number or larger size of cross-loadings may, counterintuitively, exhibit better fit. Analysis and numerical examples are used to illustrate and support the findings.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: