Pure Error REML for Analyzing Data from Multi-Stratum Designs

2025-04-23

AI Summary

In multi-tiered blocked designs (e.g., split-plot, split-split-plot), variance components lack pure error estimates, leading to biased standard errors for factorial effects. To address this, we propose a pure-error-restricted maximum likelihood (REML) estimation method grounded in the full-treatment model. This is the first approach to integrate the pure-error principle into the REML framework for multi-tiered designs, augmented with the Kenward–Roger small-sample correction. Theoretically and empirically, the method performs comparably to standard REML under correct model specification, while yielding more accurate standard error estimates and more robust hypothesis test power under response surface model misspecification. The methodology has been implemented in mainstream statistical software and validated on multiple real-world datasets.

Technology Category

• Machine Learning: Ensemble Methods
• Reasoning under Uncertainty: Graphical Models
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Since the dawn of response surface methodology, it has been recommended that designs include replicate points, so that pure error estimates of variance can be obtained and used to provide unbiased estimated standard errors of the effects of factors. In designs with more than one stratum, such as split-plot and split-split-plot designs, it is less obvious how pure error estimates of the variance components should be obtained, and no pure error estimates are given by the popular residual maximum likelihood (REML) method of estimation. We propose a method of pure error REML estimation of the variance components, using the full treatment model, obtained by treating each combination of factor levels as a discrete treatment. Our method is easy to implement using standard software and improved estimated standard errors of the fixed effects estimates can be obtained by applying the Kenward-Roger correction based on the pure error REML estimates. We illustrate the new method using several data sets and compare the performance of pure error REML with the standard REML method. The results are comparable when the assumed response surface model is correct, but the new method is considerably more robust in the case of model misspecification.

Problem

• Estimating variance components in multi-stratum designs
• Providing unbiased standard errors for factor effects
• Improving robustness against model misspecification

Innovation

• Pure error REML for multi-stratum designs
• Full treatment model with discrete treatments
• Kenward-Roger correction for improved standard errors