Issue Date
2023
Abstract
Traditional multivariate analysis of covariance test is based on restrictive assumptions like homoscedasticity and normality of errors. A semiparametric multivariate analysis of covariance model is postulated. Nonparametric regression and multivariate analysis of variance are imbedded into the backfitting algorithm to estimate the model. The responses are adjusted for covariate effect through a nonparametric function of the covariates. Simulation study indicates that a bootstrap-based test for treatment effects is correctly sized. The proposed semiparametric multivariate analysis of covariance is shown to be powerful in detecting multivariate differences in the presence of nonlinear covariates and is robust to different error distributions with homogeneous, heterogeneous, and singular variance-covariance matrices. Moreover, the test is advantageous over the classical multivariate analysis of covariance in various error variance-covariance structures especially with unbalanced sample sizes.
Source or Periodical Title
UP Los Baños Journal
Volume
21
Issue
1
Page
86-101
Document Type
Article
College
College of Arts and Sciences (CAS)
Subject
Nonparametric regression, Backfitting algorithm, Variance-covariance matrix
Recommended Citation
Almero, Leonard Allan F. and Abacan, Elfred John C., "Semiparametric multivariate analysis of covariance test of treatment effects" (2023). Journal Article. 5895.
https://www.ukdr.uplb.edu.ph/journal-articles/5895