Item Communality, Sample Size, and Variability in Latent Factor Loadings Across Simulated Samples

• In Event: Methodology and Application in Structural Equation Modeling

Abstract

This conference paper investigated the impact of item communalities and sample size in confirmatory factor analysis on the variability of factor loadings. Populations were simulated with varying item communality levels and sample sizes, and the standard deviation, average standard error, and SD-SE difference of factor loadings across trials were compared for the manipulated conditions. The results reflect prior findings that communalities and sample size influence the standard deviation and average standard error of factor loading estimates across trials and suggest that asymptotic standard errors may underestimate the standard deviation of factor loadings across trials in very small sample sizes. Practitioners should be aware that significance tests and interval estimation in CFA are strongly influenced by the investigated model features.

Authors

• Christopher Shank, Western Michigan University

• Brooks Applegate, Western Michigan University