Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2503.04775 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
Tabla de Contenidos:
- Traditional Relative Efficiency (RE), based solely on variance, has limitations in evaluating estimator performance, particularly in planned missing data designs. We introduce Bhirkuti's Relative Efficiency (BRE), a novel metric that integrates precision and accuracy to provide a more robust assessment of efficiency. To compute BRE, we use interquartile range (IQR) overlap to measure precision and apply a bias adjustment factor based on the absolute median relative bias (AMRB). Monte Carlo simulations using a Latent Growth Model (LGM) with planned missing data illustrate that BRE maintains theoretically consistency and interpretability, avoiding paradoxes such as RE exceeding 100%. Visualizations via boxplots and ridgeline plots confirm that BRE provides a stable and meaningful estimator efficiency evaluation, making it a valuable advancement in psychometric and statistical modeling. By addressing fundamental weaknesses in traditional RE, BRE provides a superior, theoretically justified alternative for relative efficiency in psychometric modeling, structural equation modeling, and missing data research. This advancement enhances data-driven decision-making and offers a methodologically rigorous tool for researchers analyzing incomplete datasets.