Doubly multivariate linear models with block exchangeable distributed errors and site-dependent covariates.

Area of Science:

• Statistics
• Multivariate Analysis
• Linear Models

Background:

• Doubly multivariate linear models are essential for analyzing complex datasets with multiple response variables.
• Testing model parameters in these settings can be computationally intensive and prone to inaccuracies.
• Existing methods like the likelihood ratio test (LRT) may not be optimal in terms of accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate Rao's score test (RST) for testing intercept and slope parameters in doubly multivariate linear models.
• To compare the performance of RST against the LRT statistic.
• To assess the computational efficiency and accuracy of the proposed RST.

Main Methods:

• Development of the RST statistic under a block exchangeable covariance structure and multivariate normality.
• Comparison of RST with LRT using Monte Carlo simulations.
• Application of the proposed method to an agricultural study involving multiple response variables from multiple trees.

Main Results:

• Rao's score test (RST) demonstrates significantly higher accuracy compared to the likelihood ratio test (LRT).
• The RST statistic requires substantially less computation time than the LRT statistic.
• Simulations confirm the robustness and efficiency of the proposed RST.

Conclusions:

• Rao's score test (RST) provides a superior method for testing parameters in doubly multivariate linear models.
• The proposed RST offers a more accurate and computationally efficient alternative for statistical analysis.
• The method is applicable to real-world scenarios, such as agricultural field studies.