A Robust Alternative Estimator for Small-Scale Latent Variable Analyses Using Categorical Indicators

• In Event: Innovative Approaches to Item Response Theory

Abstract

We extend Croon’s bias-corrected estimator to models that draw on categorical indicators (item response theory, item factor analysis). We then assess the comparative performance of the proposed estimator to conventional (marginal) ML and implement the method as an R package (BLINDED) that extends the mirt and lavaan packages. The results suggest that similar to models that use continuous indicators, Croon’s bias-corrected estimator outperforms ML in small to moderate sample sizes in terms of bias, efficiency, and power when considering categorical indicators.

Authors

• Benjamin Kelcey, University of Cincinnati

• Fangxing Bai, University of Cincinnati